From 5d43c1ce4e43798f38623b12dd278a8bd628beba Mon Sep 17 00:00:00 2001
From: Philipp Kunz <code@philkunz.com>
Date: Fri, 4 Apr 2025 12:14:41 +0000
Subject: [PATCH] fix(core): Improve PDF XML extraction, embedding, and format
detection; update loadPdf/exportPdf error handling; add new validator
implementations and enhance IPdf metadata.
---
changelog.md | 10 +
test/output/corpus-summary.md | 2 +-
test/output/test-invoice-with-xml.pdf | Bin 2283 -> 2968 bytes
test/output/xml-rechnung-corpus-results.json | 78 +--
test/output/zugferd-corpus-results.json | 224 +++----
ts/00_commitinfo_data.ts | 2 +-
ts/classes.xinvoice.ts | 45 +-
ts/formats/cii/zugferd/zugferd.encoder.ts | 633 ++++++++++++++++++-
ts/formats/factories/validator.factory.ts | 255 +++++++-
ts/formats/pdf/extractors/base.extractor.ts | 247 +++++++-
ts/formats/pdf/extractors/text.extractor.ts | 177 +++++-
ts/formats/pdf/pdf.embedder.ts | 212 ++++++-
ts/formats/pdf/pdf.extractor.ts | 118 +++-
ts/formats/utils/format.detector.ts | 355 ++++++++---
ts/interfaces/common.ts | 17 +-
15 files changed, 1957 insertions(+), 418 deletions(-)
diff --git a/changelog.md b/changelog.md
index 76fdc40..59034f6 100644
--- a/changelog.md
+++ b/changelog.md
@@ -1,5 +1,15 @@
# Changelog
+## 2025-04-04 - 4.1.6 - fix(core)
+Improve PDF XML extraction, embedding, and format detection; update loadPdf/exportPdf error handling; add new validator implementations and enhance IPdf metadata.
+
+- Update loadPdf to capture extraction result details including detected format and improve error messaging
+- Enhance TextXMLExtractor with a chunked approach using both UTF-8 and Latin-1 decoding for reliable text extraction
+- Refactor PDFEmbedder to return a structured PDFEmbedResult with proper filename normalization and robust error handling
+- Extend format detection logic by adding quickFormatCheck, isUBLFormat, isXRechnungFormat, isCIIFormat, isZUGFERDV1Format, and FatturaPA checks
+- Introduce new validator classes (UBLValidator, XRechnungValidator, FatturaPAValidator) and a generic fallback validator in ValidatorFactory
+- Update IPdf interface to include embedded XML metadata (format, filename, description) for better traceability
+
## 2025-04-03 - 4.1.5 - fix(core)
No uncommitted changes detected in the repository. The project files and functionality remain unchanged.
diff --git a/test/output/corpus-summary.md b/test/output/corpus-summary.md
index 1be8082..6276a59 100644
--- a/test/output/corpus-summary.md
+++ b/test/output/corpus-summary.md
@@ -1,6 +1,6 @@
# XInvoice Corpus Testing Summary
-Generated on: 2025-04-03T21:33:20.326Z
+Generated on: 2025-04-04T12:11:35.722Z
## Overall Summary
diff --git a/test/output/test-invoice-with-xml.pdf b/test/output/test-invoice-with-xml.pdf
index eb4203f9a4fa915e2cc2a967d96ea01c72680d33..58392ac0d24adb9238575db1d48c0f0eb89b0cea 100644
GIT binary patch
delta 1601
zcmcIk&2G~`5Eg<1+FMV(?j=%tXuRwACqY#pO=3W(v`RrBDh`glDXvm)Yj3K+#o`ev
zF1!bK-he0JCHTgU(?Cjr15QS({dQ;Oo89^L`^M+ZqH#C~c4X6MjgR;klU#5;K4rVR
z%zE+uTyg6#JW)Bbo+WY4kGX^O5r509zMiFqOV;f&mB!@4>^0whXu_#JVO0=L!AHz`
z9-gV3Q-)!2Hdb-0;vktWplb1$?*_vl5W@3d64nyJcVI5e7Q!B2Y-1c&qldZI?{a^^
z547;NdI31EU^Ug&f@(o54l^CkB9+xSeSuiWMY$=D*N1grlox9@@_Mk64{-wKgI}wM
zHE<^L0Fn>wDffbR=?}hCi${9~q9`{z6cE5NF(spo04UfJzll%>W$x>E5E{j|1`lju
zyTTEUw57CLZ?<t$nF@`*xw_*2miNc%ioYtatRT_pZ);p>LTb?U2Br?o6g6zsf+Lmd
zSr)0H>cc=|iCW=_nqDfCMB$DvI&IvG-|ld?^r3coeK0;%QMr|-!?}5UWa!}G3xKeF
z6(`}Io}=#ses0HY+HKx;WD}8~n_ik~LrN+&RHO#1XzEcCfQV5yDdqjS6ynLOquJP$
z(j<XOX6=Q!Dl53A4{m9s<0L)d)~h7#rFl{Vs~NnjjjsjC<Cip~#&Z=hE8ts`Ffy}j
zXWpEjP0>v+gIXs4C{0j;E6zCJZK}0At%1sms1E4dqIH+F4xQ2HplKawlIG4rDn9v=
zrf5$3a&Z^EV5Fv@LFeS}l5=kLrbcBcfJ=3K3RmCa?G?4$CZ%ro)1&*FUoP%EyjRRK
zHDQ8x7)UHW?zwKu<rBWdByN>Ho?c`AI%YfTn16$BN8+xkctssh3W+|cz$IpjwSdy`
z@IKf3n8`w1a#p_cN680;LadS2hSMXCH~MKh(VSk%@}5|3!pxK<+rE^n(YQN22z~(a
C5`$C#
delta 854
zcmbOs{#uYvHNeG9*HF)#OSKUQCVyo<IC&du5~s0|p|P=nMU>{`gN!2VP~K)6wrCbc
z%gH~O73xDO3sM#I{gbkSOLDpN{S?f&^xZOxic1uXjk)xFQuESFG89Y=4Y=&=xQa`P
zQWJBzD&|}bj?KSpAX0mN{f_>}iM@|4I-6H{MM^ifum!U-8Gf9gwv1(I*;I?*@9%e-
z`<b_~c<fpD<ka5z&n;}tnb>4n8YC9QG&Pt!oY2r?Tz^8LzbHbl`NQACfzB1dZ@&t(
z<ljnbuwIk0p;g~;$F{|V0<5b;KZNXF7jyMy?St70iykS|M9W@Sxa!k}%f5@Hua*@b
zdHyb8rPopk@8B&Qx@kJX8Jtq-J^3G|R80HlbJ730w#w_53fwh^ogS{3pVU%t`F7fx
z()zX&65F`=k1jW?&#;s_n9z5*!tyY8{Cvf^I}XhGG%amk`={Vsqwmw_RjsmdT_t+y
zcmIx-&|i{&@@tCrT-lmyHuZAB#>$hD{xcE``yb6Xd0ziR{)^J5Z~j;B?5xi|o1g!~
zCtgKcD15zVbzE$;?TwA=(`_}oC4YU4D*t|KYia4sjWJ5+eV(l-$Wp9dT2r<C_qx|B
zr`-AUB<`-&qX*0)yRZ31$KDk7>Uuc&$;l?~2})59RPHn!l6dX0`?8rt`<xTr^G}}_
zR*!P5uzq*vNss~8q19OzgdeJJ^Ei5CTjp80L;Y(t&s><tyx(VusMRZ}^*rM7t8exG
zeVSQlRie?w|4RNl=f$c2dlcmDue#Oqf32MN<M1Ars@{v57sRvQcSj0GT4k%xFP-8e
zz5D9jB@GXwt6Mg_nDdU8?VV}ke+IcQIZv+CycB3+0J8FvvbeYm4HOI%Kr9;@F8$!l
zs#FC-LoWTG{QMFHBcMzWm%e9STE2oYgzc7@Q<7SwpzoG5nU{qdng%0+Qq$l`&cuRC
zKU^VN!4Rn3NFf%aJtegQC}03I7|2IXj>!oLKmO0}Y@FQ4=wi^sp(b#UM=3%=l0`vi
vO%Km=NePKaa|TX6K{hsq=7}sUn2r|(rtHL`l8Pdr{YD067F?>TuKsQSjAv%_
diff --git a/test/output/xml-rechnung-corpus-results.json b/test/output/xml-rechnung-corpus-results.json
index 1aa1ad5..0d58928 100644
--- a/test/output/xml-rechnung-corpus-results.json
+++ b/test/output/xml-rechnung-corpus-results.json
@@ -1,7 +1,7 @@
{
"cii": {
- "success": 27,
- "fail": 0,
+ "success": 23,
+ "fail": 4,
"details": [
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/CII/EN16931_1_Teilrechnung.cii.xml",
@@ -137,27 +137,27 @@
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/CII/XRECHNUNG_Betriebskostenabrechnung.cii.xml",
- "success": true,
- "format": "cii",
- "error": null
+ "success": false,
+ "format": "xrechnung",
+ "error": "Wrong format detected: xrechnung, expected: cii"
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/CII/XRECHNUNG_Einfach.cii.xml",
- "success": true,
- "format": "cii",
- "error": null
+ "success": false,
+ "format": "xrechnung",
+ "error": "Wrong format detected: xrechnung, expected: cii"
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/CII/XRECHNUNG_Elektron.cii.xml",
- "success": true,
- "format": "cii",
- "error": null
+ "success": false,
+ "format": "xrechnung",
+ "error": "Wrong format detected: xrechnung, expected: cii"
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/CII/XRECHNUNG_Reisekostenabrechnung.cii.xml",
- "success": true,
- "format": "cii",
- "error": null
+ "success": false,
+ "format": "xrechnung",
+ "error": "Wrong format detected: xrechnung, expected: cii"
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/CII/not_validating_full_invoice_based_onTest_EeISI_300_CENfullmodel.cii.xml",
@@ -174,133 +174,133 @@
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_1_Teilrechnung.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_2_Teilrechnung.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_AbweichenderZahlungsempf.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Betriebskostenabrechnung.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Einfach.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Einfach_DueDate.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Einfach_negativePaymentDue.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Elektron.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_ElektronischeAdresse.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Gutschrift.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Haftpflichtversicherung_Versicherungssteuer.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Innergemeinschaftliche_Lieferungen.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Kraftfahrversicherung_Bruttopreise.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Miete.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_OEPNV.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Physiotherapeut.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Rabatte.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_RechnungsUebertragung.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Rechnungskorrektur.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Reisekostenabrechnung.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_SEPA_Prenotification.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/EN16931_Sachversicherung_berechneter_Steuersatz.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
@@ -330,13 +330,13 @@
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/not_validating_full_invoice_based_onTest_EeISI_300_CENfullmodel.ubl.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/XML-Rechnung/UBL/ubl-tc434-creditnote1.xml",
"success": true,
- "format": "xrechnung",
+ "format": "ubl",
"error": null
}
]
@@ -346,5 +346,5 @@
"fail": 0,
"details": []
},
- "totalSuccessRate": 1
+ "totalSuccessRate": 0.9272727272727272
}
\ No newline at end of file
diff --git a/test/output/zugferd-corpus-results.json b/test/output/zugferd-corpus-results.json
index e22aa24..0f2504c 100644
--- a/test/output/zugferd-corpus-results.json
+++ b/test/output/zugferd-corpus-results.json
@@ -1,13 +1,13 @@
{
"zugferdV1Correct": {
- "success": 18,
- "fail": 3,
+ "success": 21,
+ "fail": 0,
"details": [
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv1/correct/4s4u/additional-data-sample-1.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv1/correct/Intarsys/ZUGFeRD_1p0_BASIC_Einfach.pdf",
@@ -89,15 +89,15 @@
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv1/correct/Mustangproject/MustangGnuaccountingBeispielRE-20140519_499.pdf",
- "success": false,
- "format": null,
- "error": "Error: Unsupported invoice format: unknown"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv1/correct/Mustangproject/MustangGnuaccountingBeispielRE-20140522_501.pdf",
- "success": false,
- "format": null,
- "error": "Error: Unsupported invoice format: unknown"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv1/correct/Mustangproject/MustangGnuaccountingBeispielRE-20140703_502.pdf",
@@ -156,8 +156,8 @@
]
},
"zugferdV2Correct": {
- "success": 48,
- "fail": 30,
+ "success": 74,
+ "fail": 4,
"details": [
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/FNFE-factur-x-examples/Avoir_FR_type381_BASIC.pdf",
@@ -221,183 +221,183 @@
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/PHP_@gpFacturX/sample_inofficial_20190125_atgp_factur-x_v_1_0.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/BASIC/zugferd_2p0_BASIC_Einfach.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/BASIC/zugferd_2p0_BASIC_Rechnungskorrektur.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/BASIC/zugferd_2p0_BASIC_Taxifahrt.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_1_Teilrechnung.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_2_Teilrechnung.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_AbweichenderZahlungsempf.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Betriebskostenabrechnung.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Einfach.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Elektron.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_ElektronischeAdresse.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Gutschrift.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Haftpflichtversicherung_Versicherungssteuer.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Innergemeinschaftliche_Lieferungen.pdf",
"success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "format": "xrechnung",
+ "error": "Wrong format detected: xrechnung"
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Kraftfahrversicherung_Bruttopreise.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Miete.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_OEPNV.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Physiotherapeut.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Rabatte.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_RechnungsUebertragung.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Rechnungskorrektur.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Reisekostenabrechnung.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_SEPA_Prenotification.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EN16931/zugferd_2p0_EN16931_Sachversicherung_berechneter_Steuersatz.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EXTENDED/zugferd_2p0_EXTENDED_Fremdwaehrung.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EXTENDED/zugferd_2p0_EXTENDED_InnergemeinschLieferungMehrereBestellungen.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EXTENDED/zugferd_2p0_EXTENDED_Kostenrechnung.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "facturx",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EXTENDED/zugferd_2p0_EXTENDED_Rechnungskorrektur.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/EXTENDED/zugferd_2p0_EXTENDED_Warenrechnung.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/intarsys/MINIMUM/zugferd_2p0_MINIMUM.pdf",
- "success": false,
- "format": null,
- "error": "Error: No XML found in PDF"
+ "success": true,
+ "format": "zugferd",
+ "error": null
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/symtrax/Beispiele/BASIC/zugferd_2p1_BASIC_Einfach.pdf",
@@ -455,9 +455,9 @@
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/symtrax/Beispiele/EN16931/zugferd_2p1_EN16931_Betriebskostenabrechnung_XRechnung_embedded.pdf",
- "success": true,
- "format": "cii",
- "error": null
+ "success": false,
+ "format": "xrechnung",
+ "error": "Wrong format detected: xrechnung"
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/symtrax/Beispiele/EN16931/zugferd_2p1_EN16931_Einfach.pdf",
@@ -485,9 +485,9 @@
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/symtrax/Beispiele/EN16931/zugferd_2p1_EN16931_Elektron_XRechnung.pdf",
- "success": true,
- "format": "cii",
- "error": null
+ "success": false,
+ "format": "xrechnung",
+ "error": "Wrong format detected: xrechnung"
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/symtrax/Beispiele/EN16931/zugferd_2p1_EN16931_Elektron_embedded.pdf",
@@ -569,9 +569,9 @@
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/symtrax/Beispiele/EN16931/zugferd_2p1_EN16931_Reisekostenabrechnung_XRechnung_embedded.pdf",
- "success": true,
- "format": "cii",
- "error": null
+ "success": false,
+ "format": "xrechnung",
+ "error": "Wrong format detected: xrechnung"
},
{
"file": "/mnt/data/lossless/fin.cx/xinvoice/test/assets/corpus/ZUGFeRDv2/correct/symtrax/Beispiele/EN16931/zugferd_2p1_EN16931_SEPA_Prenotification.pdf",
@@ -749,5 +749,5 @@
}
]
},
- "totalCorrectSuccessRate": 0.6666666666666666
+ "totalCorrectSuccessRate": 0.9595959595959596
}
\ No newline at end of file
diff --git a/ts/00_commitinfo_data.ts b/ts/00_commitinfo_data.ts
index 2f2cf27..08da188 100644
--- a/ts/00_commitinfo_data.ts
+++ b/ts/00_commitinfo_data.ts
@@ -3,6 +3,6 @@
*/
export const commitinfo = {
name: '@fin.cx/xinvoice',
- version: '4.1.5',
+ version: '4.1.6',
description: 'A TypeScript module for creating, manipulating, and embedding XML data within PDF files specifically tailored for xinvoice packages.'
}
diff --git a/ts/classes.xinvoice.ts b/ts/classes.xinvoice.ts
index e5a41c7..7d4308b 100644
--- a/ts/classes.xinvoice.ts
+++ b/ts/classes.xinvoice.ts
@@ -189,34 +189,38 @@ export class XInvoice {
public async loadPdf(pdfBuffer: Uint8Array | Buffer, validate: boolean = false): Promise<XInvoice> {
try {
// Extract XML from PDF using the consolidated extractor
- // which tries multiple extraction methods in sequence
- const xmlContent = await this.pdfExtractor.extractXml(pdfBuffer);
-
+ const extractResult = await this.pdfExtractor.extractXml(pdfBuffer);
+
// Store the PDF buffer
this.pdf = {
name: 'invoice.pdf',
id: `invoice-${Date.now()}`,
metadata: {
- textExtraction: ''
+ textExtraction: '',
+ format: extractResult.success ? extractResult.format?.toString() : undefined
},
buffer: pdfBuffer instanceof Buffer ? new Uint8Array(pdfBuffer) : pdfBuffer
};
-
- if (!xmlContent) {
- // No XML found in PDF
- console.warn('No XML found in PDF');
- throw new Error('No XML found in PDF');
+
+ // Handle extraction result
+ if (!extractResult.success || !extractResult.xml) {
+ const errorMessage = extractResult.error ? extractResult.error.message : 'Unknown error extracting XML from PDF';
+ console.warn('XML extraction failed:', errorMessage);
+ throw new Error(`No XML found in PDF: ${errorMessage}`);
}
-
+
// Load the extracted XML
- await this.loadXml(xmlContent, validate);
-
+ await this.loadXml(extractResult.xml, validate);
+
+ // Store the detected format
+ this.detectedFormat = extractResult.format || InvoiceFormat.UNKNOWN;
+
return this;
} catch (error) {
console.error('Error loading PDF:', error);
throw error;
}
- }
+ }
/**
* Copies data from a TInvoice object
@@ -281,7 +285,7 @@ export class XInvoice {
valid: false,
errors: [{
code: 'VAL-ERROR',
- message: `Validation error: ${error.message}`
+ message: `Validation error: ${error instanceof Error ? error.message : String(error)}`
}],
level
};
@@ -356,7 +360,7 @@ export class XInvoice {
}
// Embed XML into PDF
- const modifiedPdf = await this.pdfEmbedder.createPdfWithXml(
+ const result = await this.pdfEmbedder.createPdfWithXml(
this.pdf.buffer,
xmlContent,
filename,
@@ -365,7 +369,14 @@ export class XInvoice {
this.pdf.id
);
- return modifiedPdf;
+ // Handle potential errors
+ if (!result.success || !result.pdf) {
+ const errorMessage = result.error ? result.error.message : 'Unknown error embedding XML into PDF';
+ console.error('Error exporting PDF:', errorMessage);
+ throw new Error(`Failed to export PDF: ${errorMessage}`);
+ }
+
+ return result.pdf;
}
/**
@@ -392,4 +403,4 @@ export class XInvoice {
public isFormat(format: InvoiceFormat): boolean {
return this.detectedFormat === format;
}
-}
+}
\ No newline at end of file
diff --git a/ts/formats/cii/zugferd/zugferd.encoder.ts b/ts/formats/cii/zugferd/zugferd.encoder.ts
index 5f14973..dd5c7da 100644
--- a/ts/formats/cii/zugferd/zugferd.encoder.ts
+++ b/ts/formats/cii/zugferd/zugferd.encoder.ts
@@ -2,6 +2,7 @@ import { CIIBaseEncoder } from '../cii.encoder.js';
import type { TInvoice, TCreditNote, TDebitNote } from '../../../interfaces/common.js';
import { ZUGFERD_PROFILE_IDS } from './zugferd.types.js';
import { CIIProfile } from '../cii.types.js';
+import { DOMParser, XMLSerializer } from '../../../plugins.js';
/**
* Encoder for ZUGFeRD invoice format
@@ -19,12 +20,17 @@ export class ZUGFeRDEncoder extends CIIBaseEncoder {
* @returns ZUGFeRD XML string
*/
protected async encodeCreditNote(creditNote: TCreditNote): Promise<string> {
- // Create XML root
- const xml = this.createXmlRoot();
+ // Create base XML
+ const xmlDoc = this.createBaseXml();
- // For now, return a basic XML structure
- // In a real implementation, we would populate the XML with credit note data
- return xml;
+ // Set document type code to credit note (381)
+ this.setDocumentTypeCode(xmlDoc, '381');
+
+ // Add common invoice data
+ this.addCommonInvoiceData(xmlDoc, creditNote);
+
+ // Serialize to string
+ return new XMLSerializer().serializeToString(xmlDoc);
}
/**
@@ -33,11 +39,616 @@ export class ZUGFeRDEncoder extends CIIBaseEncoder {
* @returns ZUGFeRD XML string
*/
protected async encodeDebitNote(debitNote: TDebitNote): Promise<string> {
- // Create XML root
- const xml = this.createXmlRoot();
+ // Create base XML
+ const xmlDoc = this.createBaseXml();
- // For now, return a basic XML structure
- // In a real implementation, we would populate the XML with debit note data
- return xml;
+ // Set document type code to invoice (380)
+ this.setDocumentTypeCode(xmlDoc, '380');
+
+ // Add common invoice data
+ this.addCommonInvoiceData(xmlDoc, debitNote);
+
+ // Serialize to string
+ return new XMLSerializer().serializeToString(xmlDoc);
}
-}
+
+ /**
+ * Creates a base ZUGFeRD XML document
+ * @returns XML document with basic structure
+ */
+ private createBaseXml(): Document {
+ // Create XML document from template
+ const xmlString = this.createXmlRoot();
+ const doc = new DOMParser().parseFromString(xmlString, 'application/xml');
+
+ // Add ZUGFeRD profile
+ this.addProfile(doc);
+
+ return doc;
+ }
+
+ /**
+ * Adds ZUGFeRD profile information to the XML document
+ * @param doc XML document
+ */
+ private addProfile(doc: Document): void {
+ // Get root element
+ const root = doc.documentElement;
+
+ // Create context element if it doesn't exist
+ let contextElement = root.getElementsByTagName('rsm:ExchangedDocumentContext')[0];
+ if (!contextElement) {
+ contextElement = doc.createElement('rsm:ExchangedDocumentContext');
+ root.appendChild(contextElement);
+ }
+
+ // Create guideline parameter element
+ const guidelineElement = doc.createElement('ram:GuidelineSpecifiedDocumentContextParameter');
+ contextElement.appendChild(guidelineElement);
+
+ // Add ID element with profile
+ const idElement = doc.createElement('ram:ID');
+
+ // Set profile based on the selected profile
+ let profileId = ZUGFERD_PROFILE_IDS.BASIC;
+ if (this.profile === CIIProfile.COMFORT) {
+ profileId = ZUGFERD_PROFILE_IDS.COMFORT;
+ } else if (this.profile === CIIProfile.EXTENDED) {
+ profileId = ZUGFERD_PROFILE_IDS.EXTENDED;
+ }
+
+ idElement.textContent = profileId;
+ guidelineElement.appendChild(idElement);
+ }
+
+ /**
+ * Sets the document type code in the XML document
+ * @param doc XML document
+ * @param typeCode Document type code (380 for invoice, 381 for credit note)
+ */
+ private setDocumentTypeCode(doc: Document, typeCode: string): void {
+ // Get root element
+ const root = doc.documentElement;
+
+ // Create document element if it doesn't exist
+ let documentElement = root.getElementsByTagName('rsm:ExchangedDocument')[0];
+ if (!documentElement) {
+ documentElement = doc.createElement('rsm:ExchangedDocument');
+ root.appendChild(documentElement);
+ }
+
+ // Add type code element
+ const typeCodeElement = doc.createElement('ram:TypeCode');
+ typeCodeElement.textContent = typeCode;
+ documentElement.appendChild(typeCodeElement);
+ }
+
+ /**
+ * Adds common invoice data to the XML document
+ * @param doc XML document
+ * @param invoice Invoice data
+ */
+ private addCommonInvoiceData(doc: Document, invoice: TInvoice): void {
+ // Get root element
+ const root = doc.documentElement;
+
+ // Get document element or create it
+ let documentElement = root.getElementsByTagName('rsm:ExchangedDocument')[0];
+ if (!documentElement) {
+ documentElement = doc.createElement('rsm:ExchangedDocument');
+ root.appendChild(documentElement);
+ }
+
+ // Add ID element
+ const idElement = doc.createElement('ram:ID');
+ idElement.textContent = invoice.id;
+ documentElement.appendChild(idElement);
+
+ // Add issue date element
+ const issueDateElement = doc.createElement('ram:IssueDateTime');
+ const dateStringElement = doc.createElement('udt:DateTimeString');
+ dateStringElement.setAttribute('format', '102'); // YYYYMMDD format
+ dateStringElement.textContent = this.formatDateYYYYMMDD(invoice.date);
+ issueDateElement.appendChild(dateStringElement);
+ documentElement.appendChild(issueDateElement);
+
+ // Add notes if available
+ if (invoice.notes && invoice.notes.length > 0) {
+ for (const note of invoice.notes) {
+ const noteElement = doc.createElement('ram:IncludedNote');
+ const contentElement = doc.createElement('ram:Content');
+ contentElement.textContent = note;
+ noteElement.appendChild(contentElement);
+ documentElement.appendChild(noteElement);
+ }
+ }
+
+ // Create transaction element if it doesn't exist
+ let transactionElement = root.getElementsByTagName('rsm:SupplyChainTradeTransaction')[0];
+ if (!transactionElement) {
+ transactionElement = doc.createElement('rsm:SupplyChainTradeTransaction');
+ root.appendChild(transactionElement);
+ }
+
+ // Add agreement section with seller and buyer
+ this.addAgreementSection(doc, transactionElement, invoice);
+
+ // Add delivery section
+ this.addDeliverySection(doc, transactionElement, invoice);
+
+ // Add settlement section with payment terms and totals
+ this.addSettlementSection(doc, transactionElement, invoice);
+
+ // Add line items
+ this.addLineItems(doc, transactionElement, invoice);
+ }
+
+ /**
+ * Adds agreement section with seller and buyer information
+ * @param doc XML document
+ * @param transactionElement Transaction element
+ * @param invoice Invoice data
+ */
+ private addAgreementSection(doc: Document, transactionElement: Element, invoice: TInvoice): void {
+ // Create agreement element
+ const agreementElement = doc.createElement('ram:ApplicableHeaderTradeAgreement');
+ transactionElement.appendChild(agreementElement);
+
+ // Add buyer reference if available
+ if (invoice.buyerReference) {
+ const buyerRefElement = doc.createElement('ram:BuyerReference');
+ buyerRefElement.textContent = invoice.buyerReference;
+ agreementElement.appendChild(buyerRefElement);
+ }
+
+ // Add seller
+ const sellerElement = doc.createElement('ram:SellerTradeParty');
+ this.addPartyInfo(doc, sellerElement, invoice.from);
+
+ // Add seller electronic address if available
+ if (invoice.electronicAddress && invoice.from.type === 'company') {
+ const contactElement = doc.createElement('ram:DefinedTradeContact');
+ const uriElement = doc.createElement('ram:URIID');
+ uriElement.setAttribute('schemeID', invoice.electronicAddress.scheme);
+ uriElement.textContent = invoice.electronicAddress.value;
+ contactElement.appendChild(uriElement);
+ sellerElement.appendChild(contactElement);
+ }
+
+ agreementElement.appendChild(sellerElement);
+
+ // Add buyer
+ const buyerElement = doc.createElement('ram:BuyerTradeParty');
+ this.addPartyInfo(doc, buyerElement, invoice.to);
+ agreementElement.appendChild(buyerElement);
+ }
+
+ /**
+ * Adds party information to an element
+ * @param doc XML document
+ * @param partyElement Party element
+ * @param party Party data
+ */
+ private addPartyInfo(doc: Document, partyElement: Element, party: any): void {
+ // Add name
+ const nameElement = doc.createElement('ram:Name');
+ nameElement.textContent = party.name;
+ partyElement.appendChild(nameElement);
+
+ // Add postal address
+ const addressElement = doc.createElement('ram:PostalTradeAddress');
+
+ // Add address line 1 (street)
+ if (party.address.streetName) {
+ const line1Element = doc.createElement('ram:LineOne');
+ line1Element.textContent = party.address.streetName;
+ addressElement.appendChild(line1Element);
+ }
+
+ // Add address line 2 (house number) if present
+ if (party.address.houseNumber && party.address.houseNumber !== '0') {
+ const line2Element = doc.createElement('ram:LineTwo');
+ line2Element.textContent = party.address.houseNumber;
+ addressElement.appendChild(line2Element);
+ }
+
+ // Add postal code
+ if (party.address.postalCode) {
+ const postalCodeElement = doc.createElement('ram:PostcodeCode');
+ postalCodeElement.textContent = party.address.postalCode;
+ addressElement.appendChild(postalCodeElement);
+ }
+
+ // Add city
+ if (party.address.city) {
+ const cityElement = doc.createElement('ram:CityName');
+ cityElement.textContent = party.address.city;
+ addressElement.appendChild(cityElement);
+ }
+
+ // Add country
+ if (party.address.country || party.address.countryCode) {
+ const countryElement = doc.createElement('ram:CountryID');
+ countryElement.textContent = party.address.countryCode || party.address.country;
+ addressElement.appendChild(countryElement);
+ }
+
+ partyElement.appendChild(addressElement);
+
+ // Add VAT ID if available
+ if (party.registrationDetails && party.registrationDetails.vatId) {
+ const taxRegistrationElement = doc.createElement('ram:SpecifiedTaxRegistration');
+ const taxIdElement = doc.createElement('ram:ID');
+ taxIdElement.setAttribute('schemeID', 'VA');
+ taxIdElement.textContent = party.registrationDetails.vatId;
+ taxRegistrationElement.appendChild(taxIdElement);
+ partyElement.appendChild(taxRegistrationElement);
+ }
+
+ // Add registration ID if available
+ if (party.registrationDetails && party.registrationDetails.registrationId) {
+ const regRegistrationElement = doc.createElement('ram:SpecifiedTaxRegistration');
+ const regIdElement = doc.createElement('ram:ID');
+ regIdElement.setAttribute('schemeID', 'FC');
+ regIdElement.textContent = party.registrationDetails.registrationId;
+ regRegistrationElement.appendChild(regIdElement);
+ partyElement.appendChild(regRegistrationElement);
+ }
+ }
+
+ /**
+ * Adds delivery section with delivery information
+ * @param doc XML document
+ * @param transactionElement Transaction element
+ * @param invoice Invoice data
+ */
+ private addDeliverySection(doc: Document, transactionElement: Element, invoice: TInvoice): void {
+ // Create delivery element
+ const deliveryElement = doc.createElement('ram:ApplicableHeaderTradeDelivery');
+ transactionElement.appendChild(deliveryElement);
+
+ // Add delivery date if available
+ if (invoice.deliveryDate) {
+ const deliveryDateElement = doc.createElement('ram:ActualDeliverySupplyChainEvent');
+ const occurrenceDateElement = doc.createElement('ram:OccurrenceDateTime');
+ const dateStringElement = doc.createElement('udt:DateTimeString');
+ dateStringElement.setAttribute('format', '102'); // YYYYMMDD format
+ dateStringElement.textContent = this.formatDateYYYYMMDD(invoice.deliveryDate);
+ occurrenceDateElement.appendChild(dateStringElement);
+ deliveryDateElement.appendChild(occurrenceDateElement);
+ deliveryElement.appendChild(deliveryDateElement);
+ }
+
+ // Add period of performance if available
+ if (invoice.periodOfPerformance) {
+ const periodElement = doc.createElement('ram:BillingSpecifiedPeriod');
+
+ // Start date
+ if (invoice.periodOfPerformance.from) {
+ const startDateElement = doc.createElement('ram:StartDateTime');
+ const startDateStringElement = doc.createElement('udt:DateTimeString');
+ startDateStringElement.setAttribute('format', '102'); // YYYYMMDD format
+ startDateStringElement.textContent = this.formatDateYYYYMMDD(invoice.periodOfPerformance.from);
+ startDateElement.appendChild(startDateStringElement);
+ periodElement.appendChild(startDateElement);
+ }
+
+ // End date
+ if (invoice.periodOfPerformance.to) {
+ const endDateElement = doc.createElement('ram:EndDateTime');
+ const endDateStringElement = doc.createElement('udt:DateTimeString');
+ endDateStringElement.setAttribute('format', '102'); // YYYYMMDD format
+ endDateStringElement.textContent = this.formatDateYYYYMMDD(invoice.periodOfPerformance.to);
+ endDateElement.appendChild(endDateStringElement);
+ periodElement.appendChild(endDateElement);
+ }
+
+ deliveryElement.appendChild(periodElement);
+ }
+ }
+
+ /**
+ * Adds settlement section with payment terms and totals
+ * @param doc XML document
+ * @param transactionElement Transaction element
+ * @param invoice Invoice data
+ */
+ private addSettlementSection(doc: Document, transactionElement: Element, invoice: TInvoice): void {
+ // Create settlement element
+ const settlementElement = doc.createElement('ram:ApplicableHeaderTradeSettlement');
+ transactionElement.appendChild(settlementElement);
+
+ // Add currency
+ const currencyElement = doc.createElement('ram:InvoiceCurrencyCode');
+ currencyElement.textContent = invoice.currency;
+ settlementElement.appendChild(currencyElement);
+
+ // Add payment terms
+ const paymentTermsElement = doc.createElement('ram:SpecifiedTradePaymentTerms');
+
+ // Add payment instructions if available
+ if (invoice.paymentOptions) {
+ // Add payment instructions as description - this is generic enough to work with any payment type
+ const descriptionElement = doc.createElement('ram:Description');
+ descriptionElement.textContent = `Due in ${invoice.dueInDays} days. ${invoice.paymentOptions.info || ''}`;
+ paymentTermsElement.appendChild(descriptionElement);
+ }
+
+ // Add due date
+ const dueDateElement = doc.createElement('ram:DueDateDateTime');
+ const dateStringElement = doc.createElement('udt:DateTimeString');
+ dateStringElement.setAttribute('format', '102'); // YYYYMMDD format
+
+ // Calculate due date
+ const dueDate = new Date(invoice.date);
+ dueDate.setDate(dueDate.getDate() + invoice.dueInDays);
+
+ dateStringElement.textContent = this.formatDateYYYYMMDD(dueDate.getTime());
+ dueDateElement.appendChild(dateStringElement);
+ paymentTermsElement.appendChild(dueDateElement);
+
+ settlementElement.appendChild(paymentTermsElement);
+
+ // Add payment means if available (using a generic approach)
+ if (invoice.paymentOptions) {
+ const paymentMeansElement = doc.createElement('ram:SpecifiedTradeSettlementPaymentMeans');
+
+ // Payment type code (58 for SEPA transfer as default)
+ const typeCodeElement = doc.createElement('ram:TypeCode');
+ typeCodeElement.textContent = '58';
+ paymentMeansElement.appendChild(typeCodeElement);
+
+ // Information (optional)
+ if (invoice.paymentOptions.info) {
+ const infoElement = doc.createElement('ram:Information');
+ infoElement.textContent = invoice.paymentOptions.info;
+ paymentMeansElement.appendChild(infoElement);
+ }
+
+ // If payment details are available in a standard format
+ if (invoice.paymentOptions.sepaConnection.iban) {
+ // Payee account
+ const payeeAccountElement = doc.createElement('ram:PayeePartyCreditorFinancialAccount');
+ const ibanElement = doc.createElement('ram:IBANID');
+ ibanElement.textContent = invoice.paymentOptions.sepaConnection.iban;
+ payeeAccountElement.appendChild(ibanElement);
+ paymentMeansElement.appendChild(payeeAccountElement);
+
+ // Payee financial institution if BIC available
+ if (invoice.paymentOptions.sepaConnection.bic) {
+ const institutionElement = doc.createElement('ram:PayeeSpecifiedCreditorFinancialInstitution');
+ const bicElement = doc.createElement('ram:BICID');
+ bicElement.textContent = invoice.paymentOptions.sepaConnection.bic;
+ institutionElement.appendChild(bicElement);
+ paymentMeansElement.appendChild(institutionElement);
+ }
+ }
+
+ settlementElement.appendChild(paymentMeansElement);
+ }
+
+ // Add tax details
+ this.addTaxDetails(doc, settlementElement, invoice);
+
+ // Add totals
+ this.addMonetarySummation(doc, settlementElement, invoice);
+ }
+
+ /**
+ * Adds tax details to the settlement section
+ * @param doc XML document
+ * @param settlementElement Settlement element
+ * @param invoice Invoice data
+ */
+ private addTaxDetails(doc: Document, settlementElement: Element, invoice: TInvoice): void {
+ // Calculate tax categories and totals
+ const taxCategories = new Map<number, number>(); // Map of VAT rate to net amount
+
+ // Calculate from items
+ if (invoice.items) {
+ for (const item of invoice.items) {
+ const itemNetAmount = item.unitNetPrice * item.unitQuantity;
+ const vatRate = item.vatPercentage;
+
+ const currentAmount = taxCategories.get(vatRate) || 0;
+ taxCategories.set(vatRate, currentAmount + itemNetAmount);
+ }
+ }
+
+ // Add each tax category
+ for (const [rate, baseAmount] of taxCategories.entries()) {
+ const taxElement = doc.createElement('ram:ApplicableTradeTax');
+
+ // Calculate tax amount
+ const taxAmount = baseAmount * (rate / 100);
+
+ // Add calculated amount
+ const calculatedAmountElement = doc.createElement('ram:CalculatedAmount');
+ calculatedAmountElement.textContent = taxAmount.toFixed(2);
+ taxElement.appendChild(calculatedAmountElement);
+
+ // Add type code (VAT)
+ const typeCodeElement = doc.createElement('ram:TypeCode');
+ typeCodeElement.textContent = 'VAT';
+ taxElement.appendChild(typeCodeElement);
+
+ // Add basis amount
+ const basisAmountElement = doc.createElement('ram:BasisAmount');
+ basisAmountElement.textContent = baseAmount.toFixed(2);
+ taxElement.appendChild(basisAmountElement);
+
+ // Add category code
+ const categoryCodeElement = doc.createElement('ram:CategoryCode');
+ categoryCodeElement.textContent = invoice.reverseCharge ? 'AE' : 'S';
+ taxElement.appendChild(categoryCodeElement);
+
+ // Add rate
+ const rateElement = doc.createElement('ram:RateApplicablePercent');
+ rateElement.textContent = rate.toString();
+ taxElement.appendChild(rateElement);
+
+ settlementElement.appendChild(taxElement);
+ }
+ }
+
+ /**
+ * Adds monetary summation to the settlement section
+ * @param doc XML document
+ * @param settlementElement Settlement element
+ * @param invoice Invoice data
+ */
+ private addMonetarySummation(doc: Document, settlementElement: Element, invoice: TInvoice): void {
+ const monetarySummationElement = doc.createElement('ram:SpecifiedTradeSettlementHeaderMonetarySummation');
+
+ // Calculate totals
+ let totalNetAmount = 0;
+ let totalTaxAmount = 0;
+
+ // Calculate from items
+ if (invoice.items) {
+ for (const item of invoice.items) {
+ const itemNetAmount = item.unitNetPrice * item.unitQuantity;
+ const itemTaxAmount = itemNetAmount * (item.vatPercentage / 100);
+
+ totalNetAmount += itemNetAmount;
+ totalTaxAmount += itemTaxAmount;
+ }
+ }
+
+ const totalGrossAmount = totalNetAmount + totalTaxAmount;
+
+ // Add line total amount
+ const lineTotalElement = doc.createElement('ram:LineTotalAmount');
+ lineTotalElement.textContent = totalNetAmount.toFixed(2);
+ monetarySummationElement.appendChild(lineTotalElement);
+
+ // Add tax total amount
+ const taxTotalElement = doc.createElement('ram:TaxTotalAmount');
+ taxTotalElement.textContent = totalTaxAmount.toFixed(2);
+ taxTotalElement.setAttribute('currencyID', invoice.currency);
+ monetarySummationElement.appendChild(taxTotalElement);
+
+ // Add grand total amount
+ const grandTotalElement = doc.createElement('ram:GrandTotalAmount');
+ grandTotalElement.textContent = totalGrossAmount.toFixed(2);
+ monetarySummationElement.appendChild(grandTotalElement);
+
+ // Add due payable amount
+ const duePayableElement = doc.createElement('ram:DuePayableAmount');
+ duePayableElement.textContent = totalGrossAmount.toFixed(2);
+ monetarySummationElement.appendChild(duePayableElement);
+
+ settlementElement.appendChild(monetarySummationElement);
+ }
+
+ /**
+ * Adds line items to the XML document
+ * @param doc XML document
+ * @param transactionElement Transaction element
+ * @param invoice Invoice data
+ */
+ private addLineItems(doc: Document, transactionElement: Element, invoice: TInvoice): void {
+ // Add each line item
+ if (invoice.items) {
+ for (const item of invoice.items) {
+ // Create line item element
+ const lineItemElement = doc.createElement('ram:IncludedSupplyChainTradeLineItem');
+
+ // Add line ID
+ const lineIdElement = doc.createElement('ram:AssociatedDocumentLineDocument');
+ const lineIdValueElement = doc.createElement('ram:LineID');
+ lineIdValueElement.textContent = item.position.toString();
+ lineIdElement.appendChild(lineIdValueElement);
+ lineItemElement.appendChild(lineIdElement);
+
+ // Add product information
+ const productElement = doc.createElement('ram:SpecifiedTradeProduct');
+
+ // Add name
+ const nameElement = doc.createElement('ram:Name');
+ nameElement.textContent = item.name;
+ productElement.appendChild(nameElement);
+
+ // Add article number if available
+ if (item.articleNumber) {
+ const articleNumberElement = doc.createElement('ram:SellerAssignedID');
+ articleNumberElement.textContent = item.articleNumber;
+ productElement.appendChild(articleNumberElement);
+ }
+
+ lineItemElement.appendChild(productElement);
+
+ // Add agreement information (price)
+ const agreementElement = doc.createElement('ram:SpecifiedLineTradeAgreement');
+ const priceElement = doc.createElement('ram:NetPriceProductTradePrice');
+ const chargeAmountElement = doc.createElement('ram:ChargeAmount');
+ chargeAmountElement.textContent = item.unitNetPrice.toFixed(2);
+ priceElement.appendChild(chargeAmountElement);
+ agreementElement.appendChild(priceElement);
+ lineItemElement.appendChild(agreementElement);
+
+ // Add delivery information (quantity)
+ const deliveryElement = doc.createElement('ram:SpecifiedLineTradeDelivery');
+ const quantityElement = doc.createElement('ram:BilledQuantity');
+ quantityElement.textContent = item.unitQuantity.toString();
+ quantityElement.setAttribute('unitCode', item.unitType);
+ deliveryElement.appendChild(quantityElement);
+ lineItemElement.appendChild(deliveryElement);
+
+ // Add settlement information (tax)
+ const settlementElement = doc.createElement('ram:SpecifiedLineTradeSettlement');
+
+ // Add tax information
+ const taxElement = doc.createElement('ram:ApplicableTradeTax');
+
+ // Add tax type code
+ const taxTypeCodeElement = doc.createElement('ram:TypeCode');
+ taxTypeCodeElement.textContent = 'VAT';
+ taxElement.appendChild(taxTypeCodeElement);
+
+ // Add tax category code
+ const taxCategoryCodeElement = doc.createElement('ram:CategoryCode');
+ taxCategoryCodeElement.textContent = invoice.reverseCharge ? 'AE' : 'S';
+ taxElement.appendChild(taxCategoryCodeElement);
+
+ // Add tax rate
+ const taxRateElement = doc.createElement('ram:RateApplicablePercent');
+ taxRateElement.textContent = item.vatPercentage.toString();
+ taxElement.appendChild(taxRateElement);
+
+ settlementElement.appendChild(taxElement);
+
+ // Add monetary summation
+ const monetarySummationElement = doc.createElement('ram:SpecifiedLineTradeSettlementMonetarySummation');
+
+ // Calculate item total
+ const itemNetAmount = item.unitNetPrice * item.unitQuantity;
+
+ // Add line total amount
+ const lineTotalElement = doc.createElement('ram:LineTotalAmount');
+ lineTotalElement.textContent = itemNetAmount.toFixed(2);
+ monetarySummationElement.appendChild(lineTotalElement);
+
+ settlementElement.appendChild(monetarySummationElement);
+
+ lineItemElement.appendChild(settlementElement);
+
+ // Add line item to transaction
+ transactionElement.appendChild(lineItemElement);
+ }
+ }
+ }
+
+ /**
+ * Formats a date as YYYYMMDD
+ * @param timestamp Timestamp to format
+ * @returns Formatted date string
+ */
+ private formatDateYYYYMMDD(timestamp: number): string {
+ const date = new Date(timestamp);
+ const year = date.getFullYear();
+ const month = (date.getMonth() + 1).toString().padStart(2, '0');
+ const day = date.getDate().toString().padStart(2, '0');
+ return `${year}${month}${day}`;
+ }
+}
\ No newline at end of file
diff --git a/ts/formats/factories/validator.factory.ts b/ts/formats/factories/validator.factory.ts
index beaa775..1a0b433 100644
--- a/ts/formats/factories/validator.factory.ts
+++ b/ts/formats/factories/validator.factory.ts
@@ -1,13 +1,181 @@
import { BaseValidator } from '../base/base.validator.js';
-import { InvoiceFormat } from '../../interfaces/common.js';
+import { InvoiceFormat, ValidationLevel } from '../../interfaces/common.js';
+import type { ValidationResult } from '../../interfaces/common.js';
import { FormatDetector } from '../utils/format.detector.js';
// Import specific validators
-// import { UBLValidator } from '../ubl/ubl.validator.js';
-// import { XRechnungValidator } from '../ubl/xrechnung/xrechnung.validator.js';
+import { UBLBaseValidator } from '../ubl/ubl.validator.js';
import { FacturXValidator } from '../cii/facturx/facturx.validator.js';
import { ZUGFeRDValidator } from '../cii/zugferd/zugferd.validator.js';
+/**
+ * UBL validator implementation
+ * Provides validation for standard UBL documents
+ */
+class UBLValidator extends UBLBaseValidator {
+ protected validateStructure(): boolean {
+ // Basic validation to check for required UBL invoice elements
+ if (!this.doc) return false;
+
+ let valid = true;
+
+ // Check for required UBL elements
+ const requiredElements = [
+ 'cbc:ID',
+ 'cbc:IssueDate',
+ 'cac:AccountingSupplierParty',
+ 'cac:AccountingCustomerParty'
+ ];
+
+ for (const element of requiredElements) {
+ if (!this.exists(`//${element}`)) {
+ this.addError(
+ 'UBL-STRUCT-1',
+ `Required element ${element} is missing`,
+ `/${element}`
+ );
+ valid = false;
+ }
+ }
+
+ return valid;
+ }
+
+ protected validateBusinessRules(): boolean {
+ // Basic business rule validation for UBL
+ if (!this.doc) return false;
+
+ let valid = true;
+
+ // Check that issue date is present and valid
+ const issueDateText = this.getText('//cbc:IssueDate');
+ if (!issueDateText) {
+ this.addError(
+ 'UBL-BUS-1',
+ 'Issue date is required',
+ '//cbc:IssueDate'
+ );
+ valid = false;
+ } else {
+ const issueDate = new Date(issueDateText);
+ if (isNaN(issueDate.getTime())) {
+ this.addError(
+ 'UBL-BUS-2',
+ 'Issue date is not a valid date',
+ '//cbc:IssueDate'
+ );
+ valid = false;
+ }
+ }
+
+ // Check that at least one invoice line exists
+ if (!this.exists('//cac:InvoiceLine') && !this.exists('//cac:CreditNoteLine')) {
+ this.addError(
+ 'UBL-BUS-3',
+ 'At least one invoice line or credit note line is required',
+ '/'
+ );
+ valid = false;
+ }
+
+ return valid;
+ }
+}
+
+/**
+ * XRechnung validator implementation
+ * Extends UBL validator with additional XRechnung specific validation rules
+ */
+class XRechnungValidator extends UBLValidator {
+ protected validateStructure(): boolean {
+ // Call the base UBL validation first
+ const baseValid = super.validateStructure();
+ let valid = baseValid;
+
+ // Check for XRechnung-specific elements
+ if (!this.exists('//cbc:CustomizationID[contains(text(), "xrechnung")]')) {
+ this.addError(
+ 'XRECH-STRUCT-1',
+ 'XRechnung customization ID is missing or invalid',
+ '//cbc:CustomizationID'
+ );
+ valid = false;
+ }
+
+ // Check for buyer reference which is mandatory in XRechnung
+ if (!this.exists('//cbc:BuyerReference')) {
+ this.addError(
+ 'XRECH-STRUCT-2',
+ 'BuyerReference is required in XRechnung',
+ '//'
+ );
+ valid = false;
+ }
+
+ return valid;
+ }
+
+ protected validateBusinessRules(): boolean {
+ // Call the base UBL business rule validation
+ const baseValid = super.validateBusinessRules();
+ let valid = baseValid;
+
+ // German-specific validation rules
+ // Check for proper VAT ID structure for German VAT IDs
+ const supplierVatId = this.getText('//cac:AccountingSupplierParty//cbc:CompanyID[../cac:TaxScheme/cbc:ID="VAT"]');
+ if (supplierVatId && supplierVatId.startsWith('DE') && !/^DE[0-9]{9}$/.test(supplierVatId)) {
+ this.addError(
+ 'XRECH-BUS-1',
+ 'German VAT ID format is invalid (must be DE followed by 9 digits)',
+ '//cac:AccountingSupplierParty//cbc:CompanyID'
+ );
+ valid = false;
+ }
+
+ return valid;
+ }
+}
+
+/**
+ * FatturaPA validator implementation
+ * Basic implementation for Italian electronic invoices
+ */
+class FatturaPAValidator extends BaseValidator {
+ validate(level: ValidationLevel = ValidationLevel.SYNTAX): ValidationResult {
+ // Reset errors
+ this.errors = [];
+
+ let valid = true;
+
+ if (level === ValidationLevel.SYNTAX) {
+ valid = this.validateSchema();
+ } else if (level === ValidationLevel.SEMANTIC || level === ValidationLevel.BUSINESS) {
+ valid = this.validateSchema() && this.validateBusinessRules();
+ }
+
+ return {
+ valid,
+ errors: this.errors,
+ level
+ };
+ }
+
+ protected validateSchema(): boolean {
+ // Basic schema validation for FatturaPA
+ if (!this.xml.includes('<FatturaElettronica')) {
+ this.addError('FATT-SCHEMA-1', 'Root element must be FatturaElettronica', '/');
+ return false;
+ }
+ return true;
+ }
+
+ protected validateBusinessRules(): boolean {
+ // Basic placeholder implementation - would need more detailed rules
+ // for a real implementation
+ return this.validateSchema();
+ }
+}
+
/**
* Factory to create the appropriate validator based on the XML format
*/
@@ -18,34 +186,73 @@ export class ValidatorFactory {
* @returns Appropriate validator instance
*/
public static createValidator(xml: string): BaseValidator {
- const format = FormatDetector.detectFormat(xml);
+ try {
+ const format = FormatDetector.detectFormat(xml);
- switch (format) {
- case InvoiceFormat.UBL:
- // return new UBLValidator(xml);
- throw new Error('UBL validator not yet implemented');
+ switch (format) {
+ case InvoiceFormat.UBL:
+ return new UBLValidator(xml);
- case InvoiceFormat.XRECHNUNG:
- // return new XRechnungValidator(xml);
- throw new Error('XRechnung validator not yet implemented');
+ case InvoiceFormat.XRECHNUNG:
+ return new XRechnungValidator(xml);
- case InvoiceFormat.CII:
- // For now, use Factur-X validator for generic CII
- return new FacturXValidator(xml);
+ case InvoiceFormat.CII:
+ // For now, use Factur-X validator for generic CII
+ return new FacturXValidator(xml);
- case InvoiceFormat.ZUGFERD:
- // Use dedicated ZUGFeRD validator
- return new ZUGFeRDValidator(xml);
+ case InvoiceFormat.ZUGFERD:
+ return new ZUGFeRDValidator(xml);
- case InvoiceFormat.FACTURX:
- return new FacturXValidator(xml);
+ case InvoiceFormat.FACTURX:
+ return new FacturXValidator(xml);
- case InvoiceFormat.FATTURAPA:
- // return new FatturaPAValidator(xml);
- throw new Error('FatturaPA validator not yet implemented');
+ case InvoiceFormat.FATTURAPA:
+ return new FatturaPAValidator(xml);
- default:
- throw new Error(`Unsupported invoice format: ${format}`);
+ default:
+ // For unknown formats, provide a generic validator that will
+ // mark the document as invalid but won't throw an exception
+ return new GenericValidator(xml, format);
+ }
+ } catch (error) {
+ // If an error occurs during validator creation, return a generic validator
+ // that will provide meaningful error information instead of throwing
+ console.error(`Error creating validator: ${error}`);
+ return new GenericValidator(xml, 'unknown');
}
}
}
+
+/**
+ * Generic validator for unknown or unsupported formats
+ * Provides meaningful validation errors instead of throwing exceptions
+ */
+class GenericValidator extends BaseValidator {
+ private format: string;
+
+ constructor(xml: string, format: string) {
+ super(xml);
+ this.format = format;
+ this.addError(
+ 'GEN-1',
+ `Unsupported invoice format: ${format}`,
+ '/'
+ );
+ }
+
+ validate(level: ValidationLevel = ValidationLevel.SYNTAX): ValidationResult {
+ return {
+ valid: false,
+ errors: this.errors,
+ level
+ };
+ }
+
+ protected validateSchema(): boolean {
+ return false;
+ }
+
+ protected validateBusinessRules(): boolean {
+ return false;
+ }
+}
\ No newline at end of file
diff --git a/ts/formats/pdf/extractors/base.extractor.ts b/ts/formats/pdf/extractors/base.extractor.ts
index 953a8a4..3d5f14a 100644
--- a/ts/formats/pdf/extractors/base.extractor.ts
+++ b/ts/formats/pdf/extractors/base.extractor.ts
@@ -11,7 +11,10 @@ export abstract class BaseXMLExtractor {
'factur-x.xml',
'zugferd-invoice.xml',
'ZUGFeRD-invoice.xml',
- 'xrechnung.xml'
+ 'xrechnung.xml',
+ 'ubl-invoice.xml',
+ 'invoice.xml',
+ 'metadata.xml'
];
/**
@@ -32,7 +35,8 @@ export abstract class BaseXMLExtractor {
'urn:zugferd',
'urn:factur-x',
'factur-x.eu',
- 'ZUGFeRD'
+ 'ZUGFeRD',
+ 'FatturaElettronica'
];
/**
@@ -47,7 +51,8 @@ export abstract class BaseXMLExtractor {
'</rsm:CrossIndustryDocument>',
'</ram:CrossIndustryDocument>',
'</ubl:Invoice>',
- '</ubl:CreditNote>'
+ '</ubl:CreditNote>',
+ '</FatturaElettronica>'
];
/**
@@ -69,21 +74,19 @@ export abstract class BaseXMLExtractor {
return false;
}
- // Check if it starts with XML declaration
- if (!xmlString.includes('<?xml')) {
+ // Check if it starts with XML declaration or a valid element
+ if (!xmlString.includes('<?xml') && !this.hasKnownXmlElement(xmlString)) {
return false;
}
// Check if the XML string contains known invoice formats
- const hasKnownFormat = this.knownFormats.some(format => xmlString.includes(format));
+ const hasKnownFormat = this.hasKnownFormat(xmlString);
if (!hasKnownFormat) {
return false;
}
// Check if the XML string contains binary data or invalid characters
- const invalidChars = ['\u0000', '\u0001', '\u0002', '\u0003', '\u0004', '\u0005'];
- const hasBinaryData = invalidChars.some(char => xmlString.includes(char));
- if (hasBinaryData) {
+ if (this.hasBinaryData(xmlString)) {
return false;
}
@@ -92,6 +95,11 @@ export abstract class BaseXMLExtractor {
return false;
}
+ // Check if XML has a proper structure (contains both opening and closing tags)
+ if (!this.hasProperXmlStructure(xmlString)) {
+ return false;
+ }
+
return true;
} catch (error) {
console.error('Error validating XML:', error);
@@ -99,6 +107,85 @@ export abstract class BaseXMLExtractor {
}
}
+ /**
+ * Check if the XML string contains a known element
+ * @param xmlString XML string to check
+ * @returns True if the XML contains a known element
+ */
+ protected hasKnownXmlElement(xmlString: string): boolean {
+ for (const format of this.knownFormats) {
+ // Check for opening tag of format
+ if (xmlString.includes(`<${format}`)) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ /**
+ * Check if the XML string contains a known format
+ * @param xmlString XML string to check
+ * @returns True if the XML contains a known format
+ */
+ protected hasKnownFormat(xmlString: string): boolean {
+ for (const format of this.knownFormats) {
+ if (xmlString.includes(format)) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ /**
+ * Check if the XML string has a proper structure
+ * @param xmlString XML string to check
+ * @returns True if the XML has a proper structure
+ */
+ protected hasProperXmlStructure(xmlString: string): boolean {
+ // Check for at least one matching opening and closing tag
+ for (const endTag of this.knownEndTags) {
+ const startTag = endTag.replace('/', '');
+ if (xmlString.includes(startTag) && xmlString.includes(endTag)) {
+ return true;
+ }
+ }
+
+ // If no specific tag is found but it has a basic XML structure
+ return (
+ (xmlString.includes('<?xml') && xmlString.includes('?>')) ||
+ (xmlString.match(/<[^>]+>/) !== null && xmlString.match(/<\/[^>]+>/) !== null)
+ );
+ }
+
+ /**
+ * Check if the XML string contains binary data
+ * @param xmlString XML string to check
+ * @returns True if the XML contains binary data
+ */
+ protected hasBinaryData(xmlString: string): boolean {
+ // Check for common binary data indicators
+ const binaryChars = ['\u0000', '\u0001', '\u0002', '\u0003', '\u0004', '\u0005'];
+ const consecutiveNulls = '\u0000\u0000\u0000';
+
+ // Check for control characters that shouldn't be in XML
+ if (binaryChars.some(char => xmlString.includes(char))) {
+ return true;
+ }
+
+ // Check for consecutive null bytes which indicate binary data
+ if (xmlString.includes(consecutiveNulls)) {
+ return true;
+ }
+
+ // Check for high concentration of non-printable characters
+ const nonPrintableCount = (xmlString.match(/[\x00-\x08\x0B\x0C\x0E-\x1F]/g) || []).length;
+ if (nonPrintableCount > xmlString.length * 0.05) { // More than 5% non-printable
+ return true;
+ }
+
+ return false;
+ }
+
/**
* Extract XML from a string
* @param text Text to extract XML from
@@ -108,9 +195,22 @@ export abstract class BaseXMLExtractor {
protected extractXmlFromString(text: string, startIndex: number = 0): string | null {
try {
// Find the start of the XML document
- const xmlStartIndex = text.indexOf('<?xml', startIndex);
+ let xmlStartIndex = text.indexOf('<?xml', startIndex);
+
+ // If no XML declaration, try to find known elements
if (xmlStartIndex === -1) {
- return null;
+ for (const format of this.knownFormats) {
+ const formatStartIndex = text.indexOf(`<${format.split(':').pop()}`, startIndex);
+ if (formatStartIndex !== -1) {
+ xmlStartIndex = formatStartIndex;
+ break;
+ }
+ }
+
+ // Still didn't find any start marker
+ if (xmlStartIndex === -1) {
+ return null;
+ }
}
// Try to find the end of the XML document
@@ -123,12 +223,26 @@ export abstract class BaseXMLExtractor {
}
}
+ // If no known end tag found, try to use a heuristic approach
if (xmlEndIndex === -1) {
- return null;
+ // Try to find the last closing tag
+ const lastClosingTagMatch = text.slice(xmlStartIndex).match(/<\/[^>]+>(?!.*<\/[^>]+>)/);
+ if (lastClosingTagMatch && lastClosingTagMatch.index !== undefined) {
+ xmlEndIndex = xmlStartIndex + lastClosingTagMatch.index + lastClosingTagMatch[0].length;
+ } else {
+ return null;
+ }
}
// Extract the XML content
- return text.substring(xmlStartIndex, xmlEndIndex);
+ const xmlContent = text.substring(xmlStartIndex, xmlEndIndex);
+
+ // Validate the extracted content
+ if (this.isValidXml(xmlContent)) {
+ return xmlContent;
+ }
+
+ return null;
} catch (error) {
console.error('Error extracting XML from string:', error);
return null;
@@ -143,34 +257,99 @@ export abstract class BaseXMLExtractor {
*/
protected async extractXmlFromStream(stream: PDFRawStream, fileName: string): Promise<string | null> {
try {
- // Try to decompress with pako
- const compressedBytes = stream.getContents().buffer;
+ // Get the raw bytes from the stream
+ const rawBytes = stream.getContents();
+
+ // First try without decompression (in case the content is not compressed)
+ let xmlContent = this.tryDecodeBuffer(rawBytes);
+ if (xmlContent && this.isValidXml(xmlContent)) {
+ console.log(`Successfully extracted uncompressed XML from PDF file. File name: ${fileName}`);
+ return xmlContent;
+ }
+
+ // Try with decompression
try {
- const decompressedBytes = pako.inflate(compressedBytes);
- const xmlContent = new TextDecoder('utf-8').decode(decompressedBytes);
-
- if (this.isValidXml(xmlContent)) {
- console.log(`Successfully extracted decompressed XML from PDF file. File name: ${fileName}`);
- return xmlContent;
+ const decompressedBytes = this.tryDecompress(rawBytes);
+ if (decompressedBytes) {
+ xmlContent = this.tryDecodeBuffer(decompressedBytes);
+ if (xmlContent && this.isValidXml(xmlContent)) {
+ console.log(`Successfully extracted decompressed XML from PDF file. File name: ${fileName}`);
+ return xmlContent;
+ }
}
} catch (decompressError) {
- // Decompression failed, try without decompression
- console.log(`Decompression failed for ${fileName}, trying without decompression...`);
+ console.log(`Decompression failed for ${fileName}: ${decompressError}`);
}
-
- // Try without decompression
- const rawBytes = stream.getContents();
- const rawContent = new TextDecoder('utf-8').decode(rawBytes);
-
- if (this.isValidXml(rawContent)) {
- console.log(`Successfully extracted uncompressed XML from PDF file. File name: ${fileName}`);
- return rawContent;
- }
-
+
return null;
} catch (error) {
console.error('Error extracting XML from stream:', error);
return null;
}
}
-}
+
+ /**
+ * Try to decompress a buffer using different methods
+ * @param buffer Buffer to decompress
+ * @returns Decompressed buffer or null if decompression failed
+ */
+ protected tryDecompress(buffer: Uint8Array): Uint8Array | null {
+ try {
+ // Try pako inflate (for deflate/zlib compression)
+ return pako.inflate(buffer);
+ } catch (error) {
+ // If pako fails, try other methods if needed
+ console.warn('Pako decompression failed, might be uncompressed or using a different algorithm');
+ return null;
+ }
+ }
+
+ /**
+ * Try to decode a buffer to a string using different encodings
+ * @param buffer Buffer to decode
+ * @returns Decoded string or null if decoding failed
+ */
+ protected tryDecodeBuffer(buffer: Uint8Array): string | null {
+ try {
+ // Try UTF-8 first
+ let content = new TextDecoder('utf-8').decode(buffer);
+ if (this.isPlausibleXml(content)) {
+ return content;
+ }
+
+ // Try ISO-8859-1 (Latin1)
+ content = this.decodeLatin1(buffer);
+ if (this.isPlausibleXml(content)) {
+ return content;
+ }
+
+ return null;
+ } catch (error) {
+ console.warn('Error decoding buffer:', error);
+ return null;
+ }
+ }
+
+ /**
+ * Decode a buffer using ISO-8859-1 (Latin1) encoding
+ * @param buffer Buffer to decode
+ * @returns Decoded string
+ */
+ protected decodeLatin1(buffer: Uint8Array): string {
+ return Array.from(buffer)
+ .map(byte => String.fromCharCode(byte))
+ .join('');
+ }
+
+ /**
+ * Check if a string is plausibly XML (quick check before validation)
+ * @param content String to check
+ * @returns True if the string is plausibly XML
+ */
+ protected isPlausibleXml(content: string): boolean {
+ return content.includes('<') &&
+ content.includes('>') &&
+ (content.includes('<?xml') ||
+ this.knownFormats.some(format => content.includes(format)));
+ }
+}
\ No newline at end of file
diff --git a/ts/formats/pdf/extractors/text.extractor.ts b/ts/formats/pdf/extractors/text.extractor.ts
index 8fd4731..b4a58b0 100644
--- a/ts/formats/pdf/extractors/text.extractor.ts
+++ b/ts/formats/pdf/extractors/text.extractor.ts
@@ -6,50 +6,157 @@ import { BaseXMLExtractor } from './base.extractor.js';
* Used as a fallback when other extraction methods fail
*/
export class TextXMLExtractor extends BaseXMLExtractor {
+ // Maximum chunk size to process at once (4MB)
+ private readonly CHUNK_SIZE = 4 * 1024 * 1024;
+
+ // Maximum number of chunks to check (effective 20MB search limit)
+ private readonly MAX_CHUNKS = 5;
+
+ // Common XML patterns to look for
+ private readonly XML_PATTERNS = [
+ '<?xml',
+ '<CrossIndustryInvoice',
+ '<CrossIndustryDocument',
+ '<Invoice',
+ '<CreditNote',
+ '<rsm:CrossIndustryInvoice',
+ '<rsm:CrossIndustryDocument',
+ '<ram:CrossIndustryDocument',
+ '<ubl:Invoice',
+ '<ubl:CreditNote',
+ '<FatturaElettronica'
+ ];
+
/**
* Extract XML from a PDF buffer by searching for XML patterns in the text
+ * Uses a chunked approach to handle large files efficiently
* @param pdfBuffer PDF buffer
* @returns XML content or null if not found
*/
public async extractXml(pdfBuffer: Uint8Array | Buffer): Promise<string | null> {
try {
- // Convert buffer to string and look for XML patterns
- // Increase the search range to handle larger PDFs
- const pdfString = Buffer.from(pdfBuffer).toString('utf8', 0, Math.min(pdfBuffer.length, 50000));
-
- // Look for common XML patterns in the PDF
- const xmlPatterns = [
- /<\?xml[^>]*\?>/i,
- /<CrossIndustryInvoice[^>]*>/i,
- /<CrossIndustryDocument[^>]*>/i,
- /<Invoice[^>]*>/i,
- /<CreditNote[^>]*>/i,
- /<rsm:CrossIndustryInvoice[^>]*>/i,
- /<rsm:CrossIndustryDocument[^>]*>/i,
- /<ram:CrossIndustryDocument[^>]*>/i,
- /<ubl:Invoice[^>]*>/i,
- /<ubl:CreditNote[^>]*>/i
- ];
-
- for (const pattern of xmlPatterns) {
- const match = pdfString.match(pattern);
- if (match && match.index !== undefined) {
- console.log(`Found XML pattern in PDF: ${match[0]}`);
-
- // Try to extract the XML content
- const xmlContent = this.extractXmlFromString(pdfString, match.index);
- if (xmlContent && this.isValidXml(xmlContent)) {
- console.log('Successfully extracted XML from PDF text');
- return xmlContent;
- }
- }
- }
-
- console.warn('No valid XML found in PDF text');
- return null;
+ console.log('Attempting text-based XML extraction from PDF...');
+
+ // Convert Buffer to Uint8Array if needed
+ const buffer = Buffer.isBuffer(pdfBuffer) ? new Uint8Array(pdfBuffer) : pdfBuffer;
+
+ // Try extracting XML using the chunked approach
+ return this.extractXmlFromBufferChunked(buffer);
} catch (error) {
console.error('Error in text-based extraction:', error);
return null;
}
}
-}
+
+ /**
+ * Extract XML from buffer using a chunked approach
+ * This helps avoid memory issues with large PDFs
+ * @param buffer Buffer to search in
+ * @returns XML content or null if not found
+ */
+ private extractXmlFromBufferChunked(buffer: Uint8Array): string | null {
+ // Process the PDF in chunks
+ for (let chunkIndex = 0; chunkIndex < this.MAX_CHUNKS; chunkIndex++) {
+ const startPos = chunkIndex * this.CHUNK_SIZE;
+ if (startPos >= buffer.length) break;
+
+ const endPos = Math.min(startPos + this.CHUNK_SIZE, buffer.length);
+ const chunk = buffer.slice(startPos, endPos);
+
+ // Try to extract XML from this chunk
+ const chunkResult = this.processChunk(chunk, startPos);
+ if (chunkResult) {
+ return chunkResult;
+ }
+ }
+
+ console.warn('No valid XML found in any chunk of the PDF');
+ return null;
+ }
+
+ /**
+ * Process a single chunk of the PDF buffer
+ * @param chunk Chunk buffer to process
+ * @param chunkOffset Offset position of the chunk in the original buffer
+ * @returns XML content or null if not found
+ */
+ private processChunk(chunk: Uint8Array, chunkOffset: number): string | null {
+ try {
+ // First try UTF-8 encoding for this chunk
+ const utf8String = this.decodeBufferToString(chunk, 'utf-8');
+ let xmlContent = this.searchForXmlInString(utf8String);
+
+ if (xmlContent) {
+ console.log(`Found XML content in chunk at offset ${chunkOffset} using UTF-8 encoding`);
+ return xmlContent;
+ }
+
+ // If UTF-8 fails, try Latin-1 (ISO-8859-1) which can handle binary better
+ const latin1String = this.decodeBufferToString(chunk, 'latin1');
+ xmlContent = this.searchForXmlInString(latin1String);
+
+ if (xmlContent) {
+ console.log(`Found XML content in chunk at offset ${chunkOffset} using Latin-1 encoding`);
+ return xmlContent;
+ }
+
+ // No XML found in this chunk
+ return null;
+ } catch (error) {
+ console.warn(`Error processing chunk at offset ${chunkOffset}:`, error);
+ return null;
+ }
+ }
+
+ /**
+ * Safely decode a buffer to string using the specified encoding
+ * @param buffer Buffer to decode
+ * @param encoding Encoding to use ('utf-8' or 'latin1')
+ * @returns Decoded string
+ */
+ private decodeBufferToString(buffer: Uint8Array, encoding: 'utf-8' | 'latin1'): string {
+ try {
+ if (encoding === 'utf-8') {
+ return new TextDecoder('utf-8', { fatal: false }).decode(buffer);
+ } else {
+ // For Latin-1 we can use a direct mapping (bytes 0-255 map directly to code points 0-255)
+ // This is more reliable for binary data than TextDecoder for legacy encodings
+ return Array.from(buffer)
+ .map(byte => String.fromCharCode(byte))
+ .join('');
+ }
+ } catch (error) {
+ console.warn(`Error decoding buffer using ${encoding}:`, error);
+ // Return empty string on error to allow processing to continue
+ return '';
+ }
+ }
+
+ /**
+ * Search for XML patterns in a string
+ * @param content String to search in
+ * @returns XML content or null if not found
+ */
+ private searchForXmlInString(content: string): string | null {
+ if (!content) return null;
+
+ // Search for each XML pattern
+ for (const pattern of this.XML_PATTERNS) {
+ const patternIndex = content.indexOf(pattern);
+ if (patternIndex !== -1) {
+ console.log(`Found XML pattern "${pattern}" at position ${patternIndex}`);
+
+ // Try to extract the XML content starting from the pattern position
+ const xmlContent = this.extractXmlFromString(content, patternIndex);
+
+ // Validate the extracted content
+ if (xmlContent && this.isValidXml(xmlContent)) {
+ console.log('Successfully extracted and validated XML from text');
+ return xmlContent;
+ }
+ }
+ }
+
+ return null;
+ }
+}
\ No newline at end of file
diff --git a/ts/formats/pdf/pdf.embedder.ts b/ts/formats/pdf/pdf.embedder.ts
index 8f06db2..571a34e 100644
--- a/ts/formats/pdf/pdf.embedder.ts
+++ b/ts/formats/pdf/pdf.embedder.ts
@@ -1,8 +1,33 @@
import { PDFDocument, AFRelationship } from '../../plugins.js';
import type { IPdf } from '../../interfaces/common.js';
+/**
+ * Error types for PDF embedding operations
+ */
+export enum PDFEmbedError {
+ LOAD_ERROR = 'PDF loading failed',
+ EMBED_ERROR = 'XML embedding failed',
+ SAVE_ERROR = 'PDF saving failed',
+ INVALID_INPUT = 'Invalid input parameters'
+}
+
+/**
+ * Result of a PDF embedding operation
+ */
+export interface PDFEmbedResult {
+ success: boolean;
+ data?: Uint8Array;
+ pdf?: IPdf;
+ error?: {
+ type: PDFEmbedError;
+ message: string;
+ originalError?: Error;
+ };
+}
+
/**
* Class for embedding XML into PDF files
+ * Provides robust error handling and support for different PDF formats
*/
export class PDFEmbedder {
/**
@@ -11,40 +36,92 @@ export class PDFEmbedder {
* @param xmlContent XML content to embed
* @param filename Filename for the embedded XML
* @param description Description for the embedded XML
- * @returns Modified PDF buffer
+ * @returns Result with either modified PDF buffer or error information
*/
public async embedXml(
pdfBuffer: Uint8Array | Buffer,
xmlContent: string,
filename: string = 'invoice.xml',
description: string = 'XML Invoice'
- ): Promise<Uint8Array> {
+ ): Promise<PDFEmbedResult> {
try {
+ // Validate inputs
+ if (!pdfBuffer || pdfBuffer.length === 0) {
+ return this.createErrorResult(PDFEmbedError.INVALID_INPUT, 'PDF buffer is empty or undefined');
+ }
+
+ if (!xmlContent) {
+ return this.createErrorResult(PDFEmbedError.INVALID_INPUT, 'XML content is empty or undefined');
+ }
+
+ // Ensure buffer is Uint8Array
+ const pdfBufferArray = Buffer.isBuffer(pdfBuffer) ? new Uint8Array(pdfBuffer) : pdfBuffer;
+
// Load the PDF
- const pdfDoc = await PDFDocument.load(pdfBuffer);
+ let pdfDoc: PDFDocument;
+ try {
+ pdfDoc = await PDFDocument.load(pdfBufferArray, {
+ ignoreEncryption: true, // Try to load encrypted PDFs
+ updateMetadata: false // Don't automatically update metadata
+ });
+ } catch (error) {
+ return this.createErrorResult(
+ PDFEmbedError.LOAD_ERROR,
+ `Failed to load PDF: ${error instanceof Error ? error.message : String(error)}`,
+ error instanceof Error ? error : undefined
+ );
+ }
+
+ // Normalize filename (lowercase with XML extension)
+ filename = this.normalizeFilename(filename);
// Convert the XML string to a Uint8Array
const xmlBuffer = new TextEncoder().encode(xmlContent);
- // Make sure filename is lowercase (as required by documentation)
- filename = filename.toLowerCase();
-
- // Use pdf-lib's .attach() to embed the XML
- pdfDoc.attach(xmlBuffer, filename, {
- mimeType: 'text/xml',
- description: description,
- creationDate: new Date(),
- modificationDate: new Date(),
- afRelationship: AFRelationship.Alternative,
- });
+ try {
+ // Use pdf-lib's .attach() to embed the XML
+ pdfDoc.attach(xmlBuffer, filename, {
+ mimeType: 'text/xml',
+ description: description,
+ creationDate: new Date(),
+ modificationDate: new Date(),
+ afRelationship: AFRelationship.Alternative,
+ });
+ } catch (error) {
+ return this.createErrorResult(
+ PDFEmbedError.EMBED_ERROR,
+ `Failed to embed XML: ${error instanceof Error ? error.message : String(error)}`,
+ error instanceof Error ? error : undefined
+ );
+ }
// Save the modified PDF
- const modifiedPdfBytes = await pdfDoc.save();
+ let modifiedPdfBytes: Uint8Array;
+ try {
+ modifiedPdfBytes = await pdfDoc.save({
+ addDefaultPage: false, // Don't add a page if the document is empty
+ useObjectStreams: false, // Better compatibility with older PDF readers
+ updateFieldAppearances: false // Don't update form fields
+ });
+ } catch (error) {
+ return this.createErrorResult(
+ PDFEmbedError.SAVE_ERROR,
+ `Failed to save modified PDF: ${error instanceof Error ? error.message : String(error)}`,
+ error instanceof Error ? error : undefined
+ );
+ }
- return modifiedPdfBytes;
+ return {
+ success: true,
+ data: modifiedPdfBytes
+ };
} catch (error) {
- console.error('Error embedding XML into PDF:', error);
- throw error;
+ // Catch any uncaught errors
+ return this.createErrorResult(
+ PDFEmbedError.EMBED_ERROR,
+ `Unexpected error during XML embedding: ${error instanceof Error ? error.message : String(error)}`,
+ error instanceof Error ? error : undefined
+ );
}
}
@@ -56,7 +133,7 @@ export class PDFEmbedder {
* @param description Description for the embedded XML
* @param pdfName Name for the PDF
* @param pdfId ID for the PDF
- * @returns IPdf object with embedded XML
+ * @returns Result with either IPdf object or error information
*/
public async createPdfWithXml(
pdfBuffer: Uint8Array | Buffer,
@@ -65,16 +142,101 @@ export class PDFEmbedder {
description: string = 'XML Invoice',
pdfName: string = 'invoice.pdf',
pdfId: string = `invoice-${Date.now()}`
- ): Promise<IPdf> {
- const modifiedPdfBytes = await this.embedXml(pdfBuffer, xmlContent, filename, description);
+ ): Promise<PDFEmbedResult> {
+ // Embed XML into PDF
+ const embedResult = await this.embedXml(pdfBuffer, xmlContent, filename, description);
+
+ // If embedding failed, return the error
+ if (!embedResult.success || !embedResult.data) {
+ return embedResult;
+ }
- return {
+ // Create IPdf object
+ const pdfObject: IPdf = {
name: pdfName,
id: pdfId,
metadata: {
- textExtraction: ''
+ textExtraction: '',
+ format: this.detectPdfFormat(xmlContent),
+ embeddedXml: {
+ filename: filename,
+ description: description
+ }
},
- buffer: modifiedPdfBytes
+ buffer: embedResult.data
+ };
+
+ return {
+ success: true,
+ pdf: pdfObject
};
}
-}
+
+ /**
+ * Ensures the filename is normalized according to PDF/A requirements
+ * @param filename Filename to normalize
+ * @returns Normalized filename
+ */
+ private normalizeFilename(filename: string): string {
+ // Convert to lowercase
+ let normalized = filename.toLowerCase();
+
+ // Ensure it has .xml extension
+ if (!normalized.endsWith('.xml')) {
+ normalized = normalized.replace(/\.[^/.]+$/, '') + '.xml';
+ }
+
+ // Replace invalid characters
+ normalized = normalized.replace(/[^a-z0-9_.-]/g, '_');
+
+ return normalized;
+ }
+
+ /**
+ * Tries to detect the format of the XML content
+ * @param xmlContent XML content
+ * @returns Format string or undefined
+ */
+ private detectPdfFormat(xmlContent: string): string | undefined {
+ if (xmlContent.includes('factur-x.eu') || xmlContent.includes('factur-x.xml')) {
+ return 'factur-x';
+ } else if (xmlContent.includes('zugferd') || xmlContent.includes('ZUGFeRD')) {
+ return 'zugferd';
+ } else if (xmlContent.includes('xrechnung')) {
+ return 'xrechnung';
+ } else if (xmlContent.includes('<Invoice') || xmlContent.includes('<CreditNote')) {
+ return 'ubl';
+ } else if (xmlContent.includes('FatturaElettronica')) {
+ return 'fatturapa';
+ }
+
+ return undefined;
+ }
+
+ /**
+ * Creates an error result object
+ * @param type Error type
+ * @param message Error message
+ * @param originalError Original error object
+ * @returns Error result
+ */
+ private createErrorResult(
+ type: PDFEmbedError,
+ message: string,
+ originalError?: Error
+ ): PDFEmbedResult {
+ console.error(`PDF Embedder Error (${type}): ${message}`);
+ if (originalError) {
+ console.error(originalError);
+ }
+
+ return {
+ success: false,
+ error: {
+ type,
+ message,
+ originalError
+ }
+ };
+ }
+}
\ No newline at end of file
diff --git a/ts/formats/pdf/pdf.extractor.ts b/ts/formats/pdf/pdf.extractor.ts
index 8bd243e..e7f8f1a 100644
--- a/ts/formats/pdf/pdf.extractor.ts
+++ b/ts/formats/pdf/pdf.extractor.ts
@@ -4,6 +4,32 @@ import {
AssociatedFilesExtractor,
TextXMLExtractor
} from './extractors/index.js';
+import { FormatDetector } from '../utils/format.detector.js';
+import { InvoiceFormat } from '../../interfaces/common.js';
+
+/**
+ * Error types for PDF extraction operations
+ */
+export enum PDFExtractError {
+ EXTRACT_ERROR = 'XML extraction failed',
+ INVALID_INPUT = 'Invalid input parameters',
+ NO_XML_FOUND = 'No XML found in PDF'
+}
+
+/**
+ * Result of a PDF extraction operation
+ */
+export interface PDFExtractResult {
+ success: boolean;
+ xml?: string;
+ format?: InvoiceFormat;
+ extractorUsed?: string;
+ error?: {
+ type: PDFExtractError;
+ message: string;
+ originalError?: Error;
+ };
+}
/**
* Main PDF extractor class that orchestrates the extraction process
@@ -18,9 +44,9 @@ export class PDFExtractor {
constructor() {
// Add extractors in order of preference/likelihood of success
this.extractors.push(
- new StandardXMLExtractor(), // Standard PDF/A-3 embedded files
- new AssociatedFilesExtractor(), // Associated files (ZUGFeRD v1, some Factur-X)
- new TextXMLExtractor() // Text-based extraction (fallback)
+ new StandardXMLExtractor(), // Standard PDF/A-3 embedded files
+ new AssociatedFilesExtractor(), // Associated files (ZUGFeRD v1, some Factur-X)
+ new TextXMLExtractor() // Text-based extraction (fallback)
);
}
@@ -28,36 +54,88 @@ export class PDFExtractor {
* Extract XML from a PDF buffer
* Tries multiple extraction methods in sequence
* @param pdfBuffer PDF buffer
- * @returns XML content or null if not found
+ * @returns Result with either the extracted XML or error information
*/
- public async extractXml(pdfBuffer: Uint8Array | Buffer): Promise<string | null> {
+ public async extractXml(pdfBuffer: Uint8Array | Buffer): Promise<PDFExtractResult> {
try {
console.log('Starting XML extraction from PDF...');
+ // Validate input
+ if (!pdfBuffer || pdfBuffer.length === 0) {
+ return this.createErrorResult(PDFExtractError.INVALID_INPUT, 'PDF buffer is empty or undefined');
+ }
+
+ // Ensure buffer is Uint8Array
+ const pdfBufferArray = Buffer.isBuffer(pdfBuffer) ? new Uint8Array(pdfBuffer) : pdfBuffer;
+
// Try each extractor in sequence
for (const extractor of this.extractors) {
const extractorName = extractor.constructor.name;
console.log(`Trying extraction with ${extractorName}...`);
- const xml = await extractor.extractXml(pdfBuffer);
- if (xml) {
- console.log(`Successfully extracted XML using ${extractorName}`);
- return xml;
+ try {
+ const xml = await extractor.extractXml(pdfBufferArray);
+
+ if (xml) {
+ console.log(`Successfully extracted XML using ${extractorName}`);
+
+ // Detect format of the extracted XML
+ const format = FormatDetector.detectFormat(xml);
+
+ return {
+ success: true,
+ xml,
+ format,
+ extractorUsed: extractorName
+ };
+ }
+
+ console.log(`Extraction with ${extractorName} failed, trying next method...`);
+ } catch (error) {
+ // Log error but continue with next extractor
+ console.warn(`Error using ${extractorName}: ${error instanceof Error ? error.message : String(error)}`);
}
-
- console.log(`Extraction with ${extractorName} failed, trying next method...`);
}
- // If all extractors fail, return null
- console.warn('All extraction methods failed, no valid XML found in PDF');
- return null;
+ // If all extractors fail, return a no XML found error
+ return this.createErrorResult(
+ PDFExtractError.NO_XML_FOUND,
+ 'All extraction methods failed, no valid XML found in PDF'
+ );
} catch (error) {
- console.error('Error extracting XML from PDF:', error);
- return null;
+ // Handle any unexpected errors
+ return this.createErrorResult(
+ PDFExtractError.EXTRACT_ERROR,
+ `Unexpected error during XML extraction: ${error instanceof Error ? error.message : String(error)}`,
+ error instanceof Error ? error : undefined
+ );
}
}
-
-
-
-}
+ /**
+ * Create a PDF extract result with error information
+ * @param type Error type
+ * @param message Error message
+ * @param originalError Original error object
+ * @returns Error result
+ */
+ private createErrorResult(
+ type: PDFExtractError,
+ message: string,
+ originalError?: Error
+ ): PDFExtractResult {
+ console.error(`PDF Extractor Error (${type}): ${message}`);
+ if (originalError) {
+ console.error(originalError);
+ }
+
+ return {
+ success: false,
+ error: {
+ type,
+ message,
+ originalError
+ }
+ };
+ }
+}
\ No newline at end of file
diff --git a/ts/formats/utils/format.detector.ts b/ts/formats/utils/format.detector.ts
index 4a28a60..2befe76 100644
--- a/ts/formats/utils/format.detector.ts
+++ b/ts/formats/utils/format.detector.ts
@@ -13,6 +13,18 @@ export class FormatDetector {
*/
public static detectFormat(xml: string): InvoiceFormat {
try {
+ // Quick check for empty or invalid XML
+ if (!xml || typeof xml !== 'string' || xml.trim().length === 0) {
+ return InvoiceFormat.UNKNOWN;
+ }
+
+ // Quick string-based pre-checks for performance
+ const quickCheck = FormatDetector.quickFormatCheck(xml);
+ if (quickCheck !== InvoiceFormat.UNKNOWN) {
+ return quickCheck;
+ }
+
+ // More thorough parsing-based checks
const doc = new DOMParser().parseFromString(xml, 'application/xml');
const root = doc.documentElement;
@@ -21,106 +33,26 @@ export class FormatDetector {
}
// UBL detection (Invoice or CreditNote root element)
- if (root.nodeName === 'Invoice' || root.nodeName === 'CreditNote') {
- // For simplicity, we'll treat all UBL documents as XRechnung for now
- // In a real implementation, we would check for specific customization IDs
- return InvoiceFormat.XRECHNUNG;
+ if (FormatDetector.isUBLFormat(root)) {
+ // Check for XRechnung customization
+ if (FormatDetector.isXRechnungFormat(doc)) {
+ return InvoiceFormat.XRECHNUNG;
+ }
+ return InvoiceFormat.UBL;
}
- // Factur-X/ZUGFeRD detection (CrossIndustryInvoice or CrossIndustryDocument root element)
- if (root.nodeName === 'rsm:CrossIndustryInvoice' || root.nodeName === 'CrossIndustryInvoice' ||
- root.nodeName.endsWith(':CrossIndustryInvoice')) {
- // Set up namespaces for XPath queries (ZUGFeRD v2/Factur-X)
- const namespaces = {
- rsm: 'urn:un:unece:uncefact:data:standard:CrossIndustryInvoice:100',
- ram: 'urn:un:unece:uncefact:data:standard:ReusableAggregateBusinessInformationEntity:100'
- };
-
- // Create XPath selector with namespaces
- const select = xpath.useNamespaces(namespaces);
-
- // Look for profile identifier
- const profileNode = select(
- 'string(//rsm:ExchangedDocumentContext/ram:GuidelineSpecifiedDocumentContextParameter/ram:ID)',
- doc
- );
-
- if (profileNode) {
- const profileText = profileNode.toString();
-
- // Check for ZUGFeRD profiles
- if (profileText.includes('zugferd') ||
- profileText === CII_PROFILE_IDS.ZUGFERD_BASIC ||
- profileText === CII_PROFILE_IDS.ZUGFERD_COMFORT ||
- profileText === CII_PROFILE_IDS.ZUGFERD_EXTENDED) {
- return InvoiceFormat.ZUGFERD;
- }
-
- // Check for Factur-X profiles
- if (profileText.includes('factur-x') ||
- profileText === CII_PROFILE_IDS.FACTURX_MINIMUM ||
- profileText === CII_PROFILE_IDS.FACTURX_BASIC ||
- profileText === CII_PROFILE_IDS.FACTURX_EN16931) {
- return InvoiceFormat.FACTURX;
- }
- }
-
- // If we can't determine the specific CII format, default to generic CII
- return InvoiceFormat.CII;
+ // Factur-X/ZUGFeRD detection (CrossIndustryInvoice root element)
+ if (FormatDetector.isCIIFormat(root)) {
+ return FormatDetector.detectCIIFormat(doc, xml);
}
// ZUGFeRD v1 detection (CrossIndustryDocument root element)
- if (root.nodeName === 'rsm:CrossIndustryDocument' || root.nodeName === 'CrossIndustryDocument' ||
- root.nodeName === 'ram:CrossIndustryDocument' || root.nodeName.endsWith(':CrossIndustryDocument')) {
-
- // Check for ZUGFeRD v1 namespace in the document
- const xmlString = xml.toString();
- if (xmlString.includes('urn:ferd:CrossIndustryDocument:invoice:1p0') ||
- xmlString.includes('urn:un:unece:uncefact:data:standard:ReusableAggregateBusinessInformationEntity:12') ||
- xmlString.includes('urn:ferd:CrossIndustryDocument') ||
- xmlString.includes('zugferd') ||
- xmlString.includes('ZUGFeRD')) {
- return InvoiceFormat.ZUGFERD;
- }
-
- // Set up namespaces for XPath queries (ZUGFeRD v1)
- try {
- const namespaces = {
- rsm: ZUGFERD_V1_NAMESPACES.RSM,
- ram: ZUGFERD_V1_NAMESPACES.RAM
- };
-
- // Create XPath selector with namespaces
- const select = xpath.useNamespaces(namespaces);
-
- // Look for profile identifier
- const profileNode = select(
- 'string(//rsm:SpecifiedExchangedDocumentContext/ram:GuidelineSpecifiedDocumentContextParameter/ram:ID)',
- doc
- );
-
- if (profileNode) {
- const profileText = profileNode.toString();
-
- // Check for ZUGFeRD v1 profiles
- if (profileText.includes('ferd:CrossIndustryDocument:invoice:1p0') ||
- profileText === CII_PROFILE_IDS.ZUGFERD_V1_BASIC ||
- profileText === CII_PROFILE_IDS.ZUGFERD_V1_COMFORT ||
- profileText === CII_PROFILE_IDS.ZUGFERD_V1_EXTENDED) {
- return InvoiceFormat.ZUGFERD;
- }
- }
- } catch (error) {
- console.log('Error in ZUGFeRD v1 XPath detection:', error);
- }
-
- // If we can't determine the specific profile but it's a CrossIndustryDocument, it's likely ZUGFeRD v1
+ if (FormatDetector.isZUGFeRDV1Format(root)) {
return InvoiceFormat.ZUGFERD;
}
- // FatturaPA detection would be implemented here
- if (root.nodeName === 'FatturaElettronica' ||
- (root.getAttribute('xmlns') && root.getAttribute('xmlns')!.includes('fatturapa.gov.it'))) {
+ // FatturaPA detection
+ if (FormatDetector.isFatturaPAFormat(root)) {
return InvoiceFormat.FATTURAPA;
}
@@ -130,4 +62,241 @@ export class FormatDetector {
return InvoiceFormat.UNKNOWN;
}
}
-}
+
+ /**
+ * Performs a quick format check based on string content
+ * This is faster than full XML parsing for obvious cases
+ * @param xml XML string
+ * @returns Detected format or UNKNOWN if more analysis is needed
+ */
+ private static quickFormatCheck(xml: string): InvoiceFormat {
+ const lowerXml = xml.toLowerCase();
+
+ // Check for obvious Factur-X indicators
+ if (
+ lowerXml.includes('factur-x.eu') ||
+ lowerXml.includes('factur-x.xml') ||
+ lowerXml.includes('factur-x:') ||
+ lowerXml.includes('urn:cen.eu:en16931:2017') && lowerXml.includes('factur-x')
+ ) {
+ return InvoiceFormat.FACTURX;
+ }
+
+ // Check for obvious ZUGFeRD indicators
+ if (
+ lowerXml.includes('zugferd:') ||
+ lowerXml.includes('zugferd-invoice.xml') ||
+ lowerXml.includes('urn:ferd:') ||
+ lowerXml.includes('urn:zugferd')
+ ) {
+ return InvoiceFormat.ZUGFERD;
+ }
+
+ // Check for obvious XRechnung indicators
+ if (
+ lowerXml.includes('xrechnung') ||
+ lowerXml.includes('urn:xoev-de:kosit:standard:xrechnung')
+ ) {
+ return InvoiceFormat.XRECHNUNG;
+ }
+
+ // Check for obvious FatturaPA indicators
+ if (
+ lowerXml.includes('fatturapa') ||
+ lowerXml.includes('fattura elettronica') ||
+ lowerXml.includes('fatturaelettronica')
+ ) {
+ return InvoiceFormat.FATTURAPA;
+ }
+
+ // Need more analysis
+ return InvoiceFormat.UNKNOWN;
+ }
+
+ /**
+ * Checks if the document is a UBL format
+ * @param root Root element
+ * @returns True if it's a UBL format
+ */
+ private static isUBLFormat(root: Element): boolean {
+ return (
+ root.nodeName === 'Invoice' ||
+ root.nodeName === 'CreditNote' ||
+ root.nodeName === 'ubl:Invoice' ||
+ root.nodeName === 'ubl:CreditNote' ||
+ root.nodeName.endsWith(':Invoice') ||
+ root.nodeName.endsWith(':CreditNote')
+ );
+ }
+
+ /**
+ * Checks if the document is an XRechnung format
+ * @param doc XML document
+ * @returns True if it's an XRechnung format
+ */
+ private static isXRechnungFormat(doc: Document): boolean {
+ try {
+ // Set up namespaces for XPath queries
+ const namespaces = {
+ 'cbc': 'urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-2',
+ 'ubl': 'urn:oasis:names:specification:ubl:schema:xsd:Invoice-2'
+ };
+
+ // Create XPath selector with namespaces
+ const select = xpath.useNamespaces(namespaces);
+
+ // Use getElementsByTagName directly for more reliable results
+ const customizationNodes = doc.getElementsByTagName('cbc:CustomizationID');
+
+ // Check if any CustomizationID node contains "xrechnung"
+ for (let i = 0; i < customizationNodes.length; i++) {
+ const node = customizationNodes[i];
+ if (node.textContent && node.textContent.includes('xrechnung')) {
+ return true;
+ }
+ }
+
+ return false;
+ } catch (error) {
+ console.warn('Error checking for XRechnung format:', error);
+ // If direct DOM access fails, try a string-based approach
+ const xmlStr = new XMLSerializer().serializeToString(doc);
+ return xmlStr.includes('xrechnung') || xmlStr.includes('XRechnung');
+ }
+ }
+
+ /**
+ * Checks if the document is a CII format (Factur-X/ZUGFeRD v2+)
+ * @param root Root element
+ * @returns True if it's a CII format
+ */
+ private static isCIIFormat(root: Element): boolean {
+ return (
+ root.nodeName === 'rsm:CrossIndustryInvoice' ||
+ root.nodeName === 'CrossIndustryInvoice' ||
+ root.nodeName.endsWith(':CrossIndustryInvoice')
+ );
+ }
+
+ /**
+ * Checks if the document is a ZUGFeRD v1 format
+ * @param root Root element
+ * @returns True if it's a ZUGFeRD v1 format
+ */
+ private static isZUGFeRDV1Format(root: Element): boolean {
+ return (
+ root.nodeName === 'rsm:CrossIndustryDocument' ||
+ root.nodeName === 'CrossIndustryDocument' ||
+ root.nodeName === 'ram:CrossIndustryDocument' ||
+ root.nodeName.endsWith(':CrossIndustryDocument')
+ );
+ }
+
+ /**
+ * Checks if the document is a FatturaPA format
+ * @param root Root element
+ * @returns True if it's a FatturaPA format
+ */
+ private static isFatturaPAFormat(root: Element): boolean {
+ return (
+ root.nodeName === 'FatturaElettronica' ||
+ (root.getAttribute('xmlns') && root.getAttribute('xmlns')!.includes('fatturapa.gov.it'))
+ );
+ }
+
+ /**
+ * Detects the specific CII format (Factur-X vs ZUGFeRD)
+ * @param doc XML document
+ * @param xml Original XML string for fallback checks
+ * @returns Detected format
+ */
+ private static detectCIIFormat(doc: Document, xml: string): InvoiceFormat {
+ try {
+ // Use direct DOM traversal instead of XPath for more reliable behavior
+ const contextNodes = doc.getElementsByTagNameNS(
+ 'urn:un:unece:uncefact:data:standard:CrossIndustryInvoice:100',
+ 'ExchangedDocumentContext'
+ );
+
+ if (contextNodes.length === 0) {
+ // Try without namespace
+ const noNsContextNodes = doc.getElementsByTagName('ExchangedDocumentContext');
+ if (noNsContextNodes.length === 0) {
+ // Fallback to string-based detection
+ return FormatDetector.detectCIIFormatFromString(xml);
+ }
+ }
+
+ // Loop through all potential context nodes
+ const allContextNodes = [...Array.from(contextNodes), ...Array.from(doc.getElementsByTagName('ExchangedDocumentContext'))];
+
+ for (const contextNode of allContextNodes) {
+ // Find guideline parameter
+ const guidelineNodes = contextNode.getElementsByTagName('ram:GuidelineSpecifiedDocumentContextParameter');
+
+ if (guidelineNodes.length === 0) {
+ continue;
+ }
+
+ for (const guidelineNode of Array.from(guidelineNodes)) {
+ // Find ID element
+ const idNodes = guidelineNode.getElementsByTagName('ram:ID');
+
+ if (idNodes.length === 0) {
+ continue;
+ }
+
+ for (const idNode of Array.from(idNodes)) {
+ const profileText = idNode.textContent || '';
+
+ // Check for ZUGFeRD profiles
+ if (
+ profileText.includes('zugferd') ||
+ profileText === CII_PROFILE_IDS.ZUGFERD_BASIC ||
+ profileText === CII_PROFILE_IDS.ZUGFERD_COMFORT ||
+ profileText === CII_PROFILE_IDS.ZUGFERD_EXTENDED
+ ) {
+ return InvoiceFormat.ZUGFERD;
+ }
+
+ // Check for Factur-X profiles
+ if (
+ profileText.includes('factur-x') ||
+ profileText === CII_PROFILE_IDS.FACTURX_MINIMUM ||
+ profileText === CII_PROFILE_IDS.FACTURX_BASIC ||
+ profileText === CII_PROFILE_IDS.FACTURX_EN16931
+ ) {
+ return InvoiceFormat.FACTURX;
+ }
+ }
+ }
+ }
+
+ // If we reach here, fall back to string checking
+ return FormatDetector.detectCIIFormatFromString(xml);
+ } catch (error) {
+ console.warn('Error detecting CII format, falling back to generic CII:', error);
+ return FormatDetector.detectCIIFormatFromString(xml);
+ }
+ }
+
+ /**
+ * Fallback method to detect CII format from string content
+ * @param xml XML string
+ * @returns Detected format
+ */
+ private static detectCIIFormatFromString(xml: string): InvoiceFormat {
+ // Check for Factur-X indicators
+ if (xml.includes('factur-x') || xml.includes('Factur-X')) {
+ return InvoiceFormat.FACTURX;
+ }
+
+ // Check for ZUGFeRD indicators
+ if (xml.includes('zugferd') || xml.includes('ZUGFeRD')) {
+ return InvoiceFormat.ZUGFERD;
+ }
+
+ // Generic CII if we can't determine more specifically
+ return InvoiceFormat.CII;
+ }
+}
\ No newline at end of file
diff --git a/ts/interfaces/common.ts b/ts/interfaces/common.ts
index 3633045..90e3f74 100644
--- a/ts/interfaces/common.ts
+++ b/ts/interfaces/common.ts
@@ -72,14 +72,19 @@ export interface IPdf {
id: string;
metadata: {
textExtraction: string;
+ format?: string;
+ embeddedXml?: {
+ filename: string;
+ description: string;
+ };
};
buffer: Uint8Array;
}
// Re-export types from tsclass for convenience
-export type { TInvoice } from '@tsclass/tsclass/dist_ts/finance';
-export type { TCreditNote } from '@tsclass/tsclass/dist_ts/finance';
-export type { TDebitNote } from '@tsclass/tsclass/dist_ts/finance';
-export type { TContact } from '@tsclass/tsclass/dist_ts/business';
-export type { TLetterEnvelope } from '@tsclass/tsclass/dist_ts/business';
-export type { TDocumentEnvelope } from '@tsclass/tsclass/dist_ts/business';
+export type { TInvoice } from '@tsclass/tsclass/dist_ts/finance/index.js';
+export type { TCreditNote } from '@tsclass/tsclass/dist_ts/finance/index.js';
+export type { TDebitNote } from '@tsclass/tsclass/dist_ts/finance/index.js';
+export type { TContact } from '@tsclass/tsclass/dist_ts/business/index.js';
+export type { TLetterEnvelope } from '@tsclass/tsclass/dist_ts/business/index.js';
+export type { TDocumentEnvelope } from '@tsclass/tsclass/dist_ts/business/index.js';
\ No newline at end of file