Repository Public License
1# ORIX OS Automation2# File: Workflows/Examples/02-pdf-to-structured-report.yaml3# Description: Built-in automation: extracts metadata and text statistics from the PDFs of a folder into a CSV and a PDF report.4#5# Author: Simon-Pierre Boucher6# Contact: contact@spboucher.ai7#8# Limitations9# - Text extraction works on PDFs that carry a text layer; scanned PDFs yield no text (no OCR here) and are reported10# with 0 text characters.11# - Only the first 200 000 bytes of each PDF's text are read; the character count in the report is the count of what was read.12# - The title is the PDF's own "Title" property; many PDFs have none and the column is then empty. A title containing13# a | character breaks the Markdown table (not the CSV).14# - The report is a plain table; no model is involved, so nothing is "summarised".15# - At most 100 PDFs are found per run (the default bound of filesystem.search); encrypted PDFs fail their iteration.16apiVersion: orix.ai/v117kind: Automation18metadata:19 id: pdf-to-structured-report20 name: PDF to structured data and report21 description: >-22 Reads every PDF of a folder, extracts page count, title and the size of its text layer, writes the23 result as a CSV catalogue and renders a PDF report from a Markdown table.24 tags: [pdf, documents, report]25spec:26 triggers:27 - type: manual28 parameters:29 - name: inputFolder30 type: path31 default: ~/ORIX Samples/pdf-to-structured-report/Input32 description: Folder containing the PDFs.33 - name: outputFolder34 type: path35 default: ~/ORIX Samples/pdf-to-structured-report/Output36 description: Folder that receives catalog.csv, report.md and report.pdf.37 permissions:38 filesystem:39 read: ["${{ params.inputFolder }}", "${{ params.outputFolder }}"]40 write: ["${{ params.outputFolder }}"]41 budget:42 maxDurationSeconds: 90043 maxToolCalls: 100044 steps:45 - id: find46 name: Find the PDFs47 tool: filesystem.search48 with:49 path: ${{ params.inputFolder }}50 pattern: "*.pdf"51 - id: pdfs52 needs: [find]53 transform: ${{ filter(steps.find.output.matches, m => m.type == "file") }}54 - id: prepare55 name: Create the output folder56 tool: filesystem.mkdir57 with:58 path: ${{ params.outputFolder }}59 - id: extract60 name: Read each PDF61 needs: [pdfs]62 foreach: ${{ steps.pdfs.output }}63 as: pdf64 concurrency: 265 collect: row66 steps:67 - id: info68 tool: documents.pdfInfo69 with:70 path: ${{ pdf.path }}71 - id: text72 tool: documents.pdfText73 with:74 path: ${{ pdf.path }}75 maxBytes: 20000076 - id: row77 needs: [info, text]78 transform: ${{ [basename(pdf.path), steps.info.output.pageCount, default(steps.info.output.title, ""), sum(steps.text.output.pages, p => len(p.text))] }}79 - id: catalog80 name: Write the CSV catalogue81 needs: [extract, prepare]82 tool: documents.csvWrite83 with:84 path: ${{ params.outputFolder }}/catalog.csv85 columns: [file, pages, title, textCharacters]86 rows: ${{ steps.extract.output.results }}87 overwrite: true88 - id: compose89 name: Compose the report90 needs: [extract]91 transform: |92 ${{ "# PDF report\n\n"93 + "Folder: " + params.inputFolder + "\n\n"94 + "Generated: " + env.DATETIME + "\n\n"95 + "Documents: " + steps.extract.output.count + " — pages in total: " + sum(steps.extract.output.results, r => r[1]) + "\n\n"96 + "| File | Pages | Title | Text characters |\n|---|---|---|---|\n"97 + join(map(steps.extract.output.results, r => "| " + join(r, " | ") + " |"), "\n") + "\n" }}98 - id: markdown99 name: Write the Markdown report100 needs: [compose, prepare]101 tool: filesystem.write102 with:103 path: ${{ params.outputFolder }}/report.md104 overwrite: true105 content: ${{ steps.compose.output }}106 - id: render107 name: Render the report as PDF108 needs: [compose, prepare]109 tool: documents.renderPDF110 with:111 content: ${{ steps.compose.output }}112 destination: ${{ params.outputFolder }}/report.pdf113 overwrite: true114 - id: check115 name: Verify the artifacts116 needs: [catalog, markdown, render]117 verify:118 - truthy: ${{ steps.extract.output.count > 0 }}119 - equals: ["${{ steps.catalog.output.rows }}", "${{ steps.extract.output.count }}"]120 - fileNonEmpty: ${{ params.outputFolder }}/catalog.csv121 - fileContains: ${{ params.outputFolder }}/report.md122 text: "| File | Pages |"123 - fileNonEmpty: ${{ params.outputFolder }}/report.pdf124 - truthy: ${{ steps.render.output.pages >= 1 }}125 - id: result126 needs: [check]127 output:128 documents: ${{ steps.extract.output.count }}129 pages: ${{ sum(steps.extract.output.results, r => r[1]) }}130 catalog: ${{ params.outputFolder }}/catalog.csv131 report: ${{ params.outputFolder }}/report.pdf132 artifacts:133 - ${{ params.outputFolder }}/catalog.csv134 - ${{ params.outputFolder }}/report.pdf135