]+src="([^"]+)"', re.IGNORECASE)
img_urls = img_pattern.findall(html_content)
image_map = {} # old URL → local path
for i, img_url in enumerate(img_urls, 1):
try:
# Google Docs images are served from googleusercontent.com
if not img_url.startswith("http"):
continue
req = urllib.request.Request(img_url, headers={"User-Agent": "PostmanDevRelSkill/1.0"})
img_data = urllib.request.urlopen(req, timeout=30).read()
# Determine extension from content or default to png
ext = "png"
if img_url.endswith(".jpg") or img_url.endswith(".jpeg"):
ext = "jpg"
elif img_url.endswith(".gif"):
ext = "gif"
elif img_url.endswith(".svg"):
ext = "svg"
filename = f"image-{i}.{ext}"
filepath = os.path.join(image_dir, filename)
with open(filepath, "wb") as f:
f.write(img_data)
image_map[img_url] = f"images/{slug}/{filename}"
print(f"Downloaded: {filename} ({len(img_data)} bytes)")
except Exception as e:
print(f"Failed to download image {i}: {e}")
```
### Step 4: Convert HTML to Markdown
Write and run a Python script to convert the HTML content to clean markdown:
```python
import subprocess, sys
# Install html2text if needed
subprocess.check_call([sys.executable, "-m", "pip", "install", "html2text", "-q"])
import html2text
h = html2text.HTML2Text()
h.body_width = 0 # Don't wrap lines
h.ignore_links = False
h.ignore_images = False
h.ignore_emphasis = False
h.protect_links = True
h.unicode_snob = True
h.skip_internal_links = True
markdown = h.handle(html_content)
```
### Step 5: Clean Up the Markdown
After conversion, clean up the markdown:
1. **Replace image URLs** — swap the Google-hosted image URLs with local paths from `image_map`
2. **Remove Google Docs artifacts** — strip empty links, tracking parameters, Google redirect wrappers
3. **Fix heading levels** — ensure the document starts with H1 and has proper hierarchy
4. **Clean up whitespace** — remove excessive blank lines (more than 2 consecutive)
5. **Add language identifiers** to code blocks that are missing them
```python
# Replace image URLs with local paths
for old_url, local_path in image_map.items():
markdown = markdown.replace(old_url, local_path)
# Unwrap Google redirect links: https://www.google.com/url?q=REAL_URL&...
google_redirect = re.compile(r'https://www\.google\.com/url\?q=(.*?)&[^)]*')
markdown = google_redirect.sub(lambda m: urllib.parse.unquote(m.group(1)), markdown)
# Remove excessive blank lines
markdown = re.sub(r'\n{3,}', '\n\n', markdown)
# Strip Google Docs comment anchors and bookmark links
markdown = re.sub(r'\[([^\]]*)\]\(#[^)]*\)', r'\1', markdown)
```
### Step 6: Add Blog Frontmatter
Add YAML frontmatter at the top of the markdown file with metadata extracted from the document:
```yaml
---
suggested_title: "Title from the Google Doc"
meta_description: "Generated from the first paragraph — under 155 characters"
seo_score: 0
primary_keyword: ""
secondary_keywords: []
source_gdoc: "https://docs.google.com/document/d/{doc_id}/edit"
---
```
Set `seo_score` to 0 and leave keyword fields empty — the `blog-copyeditor` skill will fill these in during the copyedit pass.
Extract the `meta_description` from the first paragraph of the document body (under 155 characters, ending at a natural sentence boundary).
### Step 7: Save the Markdown File
Save the final markdown to `blog-output/{slug}.md`:
```python
output_path = f"blog-output/{slug}.md"
with open(output_path, "w") as f:
f.write(markdown)
print(f"Saved: {output_path}")
```
### Step 8: Present Results
Summarize what was created:
```
Google Doc converted to markdown.
Source: https://docs.google.com/document/d/{doc_id}/edit
Title: "Document Title Here"
Output: blog-output/{slug}.md
Images: 3 images saved to blog-output/images/{slug}/
Next steps:
- Run /blog-copyeditor blog-output/{slug}.md to copyedit and add SEO metadata
- Run /blog-header-image blog-output/{slug}.md to generate a header image
- Run /blog-wordpress-stage blog-output/{slug}.md to stage to WordPress
```
## Error Handling
- **Invalid URL:** If the URL doesn't contain a Google Docs document ID, tell the user and show the expected format.
- **Access denied (403):** The document is private. Tell the user to change sharing to "Anyone with the link can view".
- **Document not found (404):** The document ID is invalid or the doc was deleted.
- **Image download failure:** Continue with the conversion — note which images failed and use the original URL as a placeholder in the markdown.
- **html2text not available:** Fall back to a basic regex-based HTML-to-markdown conversion (strip tags, convert `` to `#`, `
` to paragraphs, etc.).
## Important Guidelines
- **Do not modify the Google Doc** — this skill only reads from it.
- **Preserve all content** — tables, lists, code blocks, links, and formatting should survive the conversion.
- **Download all images locally** — don't leave references to Google-hosted URLs, as they expire.
- **Use the document title as the filename** — slugified, lowercase, hyphenated.
- **Add frontmatter with empty SEO fields** — the copyeditor skill fills these in later.
- **Unwrap Google redirect URLs** — Google Docs wraps all links in `google.com/url?q=` redirects. Always unwrap to the original URL.