Initial Commit

action.yml

@@ -0,0 +1,29 @@
+name: PDF To Markdown workflow
+description: Updates README.md generating PDF images
+author: northy
+
+inputs:
+  pdf_input:
+    description: 'Input PDF file'
+    default: 'source.pdf'
+    required: false
+  output_folder:
+    description: 'Output pngs folder'
+    default: 'output_pngs'
+    required: false
+  source_md:
+    description: "Source markdown file"
+    default: 'source.md'
+    required: false
+  output_md:
+    description: "Output markdown file"
+    default: 'README.md'
+    required: false
+
+branding:
+  icon: 'activity'
+  color: 'yellow'
+
+runs:
+  using: 'node12'
+  main: 'run.js'

requirements.txt

@@ -0,0 +1,2 @@
+pymupdf
+pydotenv

run.js

@@ -0,0 +1,26 @@
+const spawn = require('child_process').spawn;
+const path = require("path");
+
+const exec = (cmd, args=[]) => new Promise((resolve, reject) => {
+    console.log(`Started: ${cmd} ${args.join(" ")}`)
+    const app = spawn(cmd, args, { stdio: 'inherit' });
+    app.on('close', code => {
+        if(code !== 0){
+            err = new Error(`Invalid status code: ${code}`);
+            err.code = code;
+            return reject(err);
+        };
+        return resolve(code);
+    });
+    app.on('error', reject);
+});
+
+const main = async () => {
+    await exec('bash', [path.join(__dirname, './run.sh')]);
+};
+
+main().catch(err => {
+    console.error(err);
+    console.error(err.stack);
+    process.exit(err.code || -1);
+})

run.sh

@@ -0,0 +1,17 @@
+#!/bin/sh
+set -e
+
+wget -O split.py https://raw.githubusercontent.com/northy/pdf-to-markdown-workflow/master/split.py
+
+python split.py ${INPUT_PDF_INPUT} ${INPUT_OUTPUT_FOLDER} ${INPUT_SOURCE_MD} ${INPUT_OUTPUT_MD}
+
+rm split.py
+
+git config --local user.email "[email protected]"
+git config --local user.name "GitHub Action"
+git add ${INPUT_OUTPUT_FOLDER} ${INPUT_OUTPUT_MD}
+git commit -m "Compile MD"
+
+remote_repo="https://${GITHUB_ACTOR}:${GITHUB_TOKEN}@github.com/${GITHUB_REPOSITORY}.git"
+
+git push "${remote_repo}";

split.py

@@ -0,0 +1,42 @@
+import fitz, sys, os
+
+pdf_input = sys.argv[1]
+output_folder = sys.argv[2]
+source_md = sys.argv[3]
+output_md = sys.argv[4]
+
+if not os.path.exists(output_folder):
+    os.makedirs(output_folder)
+
+doc = fitz.open(pdf_input)
+source = open(source_md,'r',encoding='utf-8')
+lines = source.readlines()
+source.close()
+
+s = 0
+e = 0
+
+for i in range(len(lines)) :
+    if lines[i].find("<!-- PDF-TO-MARKDOWN:START -->")!=-1 : s = i
+    if lines[i].find("<!-- PDF-TO-MARKDOWN:END -->")!=-1 : e = i
+for i in range(e-s-1) :
+    del lines[s+1]
+
+i=1
+imgs = []
+while True :
+    try :
+        page = doc.loadPage(i-1)
+        pix = page.getPixmap()
+        output = f"{output_folder}/page{i}.png"
+        imgs.append(f'![Page {i}]({output} "Page {i}")\n---\n')
+        pix.writePNG(output)
+    except :
+        break
+    i+=1
+
+lines = lines[:s+1]+imgs+lines[s+1:]
+
+md = open(output_md,'w',encoding='utf-8')
+md.writelines(lines)
+md.close()