GPT2Org

This extension is compatible with Google Chrome and Firefox and introduces a “Summary” icon to the user interface. When clicked, it will extract the main textual content from the entire web page using the standalone version of the readability library, which is also utilized in the [Firefox Reader View](https://support.mozilla.org/kb/firefox-reader-view-clutter-free-web-pages) View feature. The extracted text is then processed by [OpenAI](https://openai.com/)’s Language Model (LLM) to generate a concise summary. Users have the option to customize the type of operation performed on the extracted text by providing a specific prompt. The summarized content can be saved via org-protocol, with detailed instructions available on the [Org-Mode](https://orgmode.org/) website for configuring this functionality. The extension includes four pre-configured capture buttons that correspond to different org-protocol templates. These templates enable users to save the summarized content as a new [org-roam](https://www.orgroam.com/) node, append it to an existing [org-roam](https://www.orgroam.com/) node, track it as a [clocking](https://orgmode.org/manual/Clocking-commands.html) item, or add it to their daily journal.

Utilizing the OpenAI API, one can efficiently extract a succinct summary from a webpage by providing the necessary prompt and API key. This API facilitates the generation of a condensed overview of the webpage’s content. Subsequently, the obtained summary can be organized and stored in a structured format, such as an Org file within the Emacs environment or an Org-roam node by org protocol. By leveraging artificial intelligence technology and the organizational capabilities of the org-mode, individuals can enhance the process of summarizing information sourced from the web and seamlessly integrate it into their personal knowledge management systems.

Getting started

Set your org or org-roam templates

Templates:

(setq org-capture-templates
      '(("orp" "Org roam capture content" plain (file (lambda () (org-roam-node-file (org-roam-node-read))))
         "* [[%:link][%(my-org/transform-square-brackets-to-round-ones \"%:description\")]]\n - Source: %u\n#+BEGIN_QUOTE\n%(my-org/format-i-string)\n#+END_QUOTE")
        ("orc" "Org roam capture content to clocked" plain (clock)
         "* [[%:link][%(my-org/transform-square-brackets-to-round-ones \"%:description\")]]\n - Source: %u\n#+BEGIN_QUOTE\n%(my-org/format-i-string)\n#+END_QUOTE")
        ("orj" "Org roam capture content to journal" entry (file+olp+datetree "~/Dropbox/org-roam/20241007093145-diary.org")
         "* [[%:link][%(my-org/transform-square-brackets-to-round-ones \"%:description\")]]\n - Source: %u\n#+BEGIN_QUOTE\n%(my-org/format-i-string)\n#+END_QUOTE")))
(setq org-roam-capture-ref-templates
      '(("p" "Protocol" plain
         "* [[${ref}][%(my-org/transform-square-brackets-to-round-ones \"${title}\")]]\n - Source: %u\n#+BEGIN_QUOTE\n%(my-org/format-i-string)\n#+END_QUOTE"
         :target (file+head "${slug}.org" "#+title: ${title}")
         :unnarrowed t)))

Functions used in above templates

(defun my-org/transform-square-brackets-to-round-ones(string-to-transform)
  "Transforms [ into ( and ] into ), other chars left unchanged."
  (concat
   (mapcar #'(lambda (c) (if (equal c ?\[) ?\( (if (equal c ?\]) ?\) c))) string-to-transform)))

(defun my-org/replace-in-string (what with in)
  (replace-regexp-in-string (regexp-quote what) with in nil 'literal))

(defun my-org/format-i-string ()
  (let ((v-i
         (my-org/replace-in-string
          "\"" "\\\""
          (org-link-unescape
           (plist-get org-store-link-plist :initial)))))
    (if (not (or (string-empty-p v-i)    ; Check if quote-str is an empty string
                 (string-match-p "^\\s*$" v-i)))  ; Check if quote-str is only whitespace
        (if (yes-or-no-p "Format the quote?")
            (shell-command-to-string (format "~/miniconda3/bin/python3 ~/bin/pfmt.py --string \"%s\"" v-i))
          v-i)
      "")))

The format python script

#!/opt/homebrew/bin/python3

import argparse
import cjkwrap

from itertools import groupby

def paragraph(lines) :
    for group_separator, line_iteration in groupby(lines.splitlines(True),
                                                   key = str.isspace) :
        if not group_separator :
            yield ''.join(line_iteration)


def main():
    parser = argparse.ArgumentParser(description = 'format string vector')
    parser.add_argument('--string', dest='inString', help='input string')
    args = parser.parse_args()

    inString = args.inString.strip(" \n")
    reString = ""

    for p in paragraph(inString):
        p = p.strip(" \n")
        reString = "{0}\n\n{1}".format(reString, cjkwrap.fill(p, 81))

    reString = reString.strip("\n")
    print(reString)

if __name__ == '__main__':
    main()

Seting org-protocol

See README_old.md .

Acknowedgement

This project referred, forked, or used some parts of the codes from the other projects:

Project URL	Usage	Licenses of Used Parts
org-capture-extension	Copy some javascript code	MIT
org-roam-capture-extension	As the initial framework	MIT
Readability.js	To extract the main text from web page	Apache License 2.0