This program extracts annotations (highlights, comments, etc.) from a PDF file, and formats them as Markdown or exports them to JSON. It is primarily intended for use in reviewing submissions to scientific conferences/journals.

For the default Markdown format, the output is as follows:

• Highlights without an attached comment are output first, as "highlights" with just the highlighted text included. Note that these are not typically suitable for use in a review, since they're unlikely to have any meaning to the recipient; they are just meant to serve as a reminder to the reviewer.

• Highlights with an attached comment, and text annotations (not attached to any particular text/highlight) are output next, as "detailed comments". Typically most comments on a reviewed paper are of this form.

• Underline, strikeout, and squiggly underline annotations are output last, as "Nits", with or without an attached comment. The intention of this is to easily separate formatting or grammatical corrections from more substantial comments about the content of the document.

For each annotation, the page number is given, along with the associated (highlighted/underlined) text, if any. Additionally, if the document embeds outlines (aka bookmarks), such as those generated by the LaTeX hyperref package, they are printed to help identify to which section in the document the annotation refers.

See pdfannots --help (in a source tree: pdfannots.py --help) for options and invocation.

• Python >= 3.7
• pdfminer.six

• While it is generally reliable, pdfminer (the underlying PDF parser) is not infallible at extracting text from a PDF. It has been known to fail in several different ways:

  • Sometimes it misses or misplaces individual characters, resulting in annotations with some or all of the text missing (in the latter case, you'll see a warning).

  • Sometimes the characters are captured, but not spaces between the words. Tweaking the advanced layout analysis parameters (e.g., --word-margin) may help with this.

  • Sometimes it extracts all the text but renders it out of order, for example, reporting that text at the top of a second column comes before text at the end of the first column. This causes pdfannots to return the annotations out of order, or to report the wrong outlines (section headings) for annotations. You can mostly work around this issue by using the --cols parameter to force a fixed page layout for the document (e.g. --cols=2 for a typical 2-column document).

• If an annotation (such as a StrikeOut) covers solely whitespace, no text is extracted for the annotation, and it will be skipped (with a warning). This is an artifact of the way pdfminer reports whitespace with only an implicit position defined by surrounding characters.

• When extracting text, we remove all hyphens that immediately precede a line break and join the adjacent words. This usually produces the best results with LaTeX multi-column documents (e.g. "soft-\nware" becomes "software"), but sometimes the hyphen needs to stay (e.g. "memory-\nmapped", which will be extracted as "memorymapped"), and we can't tell the difference. To disable this behaviour, pass --keep-hyphens.

1. I'd like to change how the output is formatted.

   Some minor tweaks (e.g.: word wrap, skipping or reordering output sections) can be accomplished via command-line arguments.

   All of the output comes from the relevant Printer subclass; more elaborate changes can be accomplished there. Pull requests to introduce new output formats or variants as printers are welcomed.

2. I think I got a review generated by this tool...

   I hope that it was a constructive review, and that the annotations helped the reviewer give you more detailed feedback so you can improve your paper. This is, after all, just a tool, and it should not be an excuse for reviewer sloppiness.