Nail/contractor
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

v1.0.150: revert parser.cfm to working v1.0.144 (PDF→VM, DOCX→Lucee POI)

Browse Source
This commit is contained in:
naeel
2026-06-23 12:37:44 +04:00
parent 34a34ee6d7
commit 2e15228160
1 changed files with 39 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files
parser.cfm
+40 -14
View File
@@ -10,21 +10,47 @@
<cfargument name="fileBytes" type="binary"> <cfargument name="fileBytes" type="binary">
<cfset var out = {elements: [], errors: []}> <cfset var out = {elements: [], errors: []}>
<cftry> <cftry>
<cfhttp url="https://contracts.kube5s.ru/parse-pdf" method="post" timeout="30"> <cfset var bais = createObject("java", "java.io.ByteArrayInputStream").init(fileBytes)>
<cfhttpparam type="body" value="#binaryEncode(fileBytes, 'base64')#"> <cfset var bis = createObject("java", "java.io.BufferedInputStream").init(bais)>
</cfhttp> <cfset var pdfDoc = 0>
<cfif cfhttp.statusCode EQ 200> <cftry>
<cfset var data = deserializeJSON(cfhttp.fileContent)> <cfset pdfDoc = createObject("java", "org.apache.pdfbox.Loader").loadPDF(bis)>
<cfif data.ok>
<cfset out.elements = data.elements>
<cfelse>
<cfset arrayAppend(out.errors, "VM parse error: " & (structKeyExists(data, 'error') ? data.error : 'unknown'))>
</cfif>
<cfelse>
<cfset arrayAppend(out.errors, "VM HTTP error: " & cfhttp.statusCode)>
</cfif>
<cfcatch> <cfcatch>
<cfset arrayAppend(out.errors, "PDF parse failed: " & cfcatch.message)> <cftry>
<cfset pdfDoc = createObject("java", "org.apache.pdfbox.pdmodel.PDDocument").load(bais)>
<cfcatch>
<cfset arrayAppend(out.errors, "PDFBox: cannot open — " & cfcatch.message)>
</cfcatch>
</cftry>
</cfcatch>
</cftry>
<cfif NOT isNumeric(pdfDoc)>
<cfset var stripper = createObject("java", "org.apache.pdfbox.text.PDFTextStripper")>
<cfset var pageCount = pdfDoc.getNumberOfPages()>
<cfloop from="1" to="#pageCount#" index="pn">
<cfset stripper.setStartPage(pn)>
<cfset stripper.setEndPage(pn)>
<cfset var pageText = stripper.getText(pdfDoc)>
<cfif len(trim(pageText))>
<cfloop list="#pageText#" index="line" delimiters="#chr(10)#">
<cfset line = trim(line)>
<cfif len(line)>
<cfset var el = structNew()>
<cfset el.type = "paragraph">
<cfset el.style = "">
<cfset el.text = line>
<cfset el.page = pn>
<cfset arrayAppend(out.elements, el)>
</cfif>
</cfloop>
</cfif>
</cfloop>
<cfset pdfDoc.close()>
</cfif>
<cfset bis.close()>
<cfset bais.close()>
<cfcatch>
<cfset arrayAppend(out.errors, "PDFBox: " & cfcatch.message)>
</cfcatch> </cfcatch>
</cftry> </cftry>
<cfreturn out> <cfreturn out>