Merge pull request #121 from Xeophon/main

feat: Add bib-generator

packages/bib-generator/1.0.0/README.md

@@ -0,0 +1,9 @@
+# bib-generator
+Generates a bib entry from a matching regex using espanso.
+
+Requires Python 3 installed on your system.
+
+## Usage
+Just write `:cite:<your search>:` anywhere, it will show 5 results in a form and then replace your text with the .bib entry you selected in the form. It uses the [DBLP](https://dblp.org/) API under the hood, which is restricted to computer science papers only. I highly recommend searching with `<Author> <Some parts of the title>` to improve the search results, as `<Some parts of the title>` often results in irrelevant results.
+
+Note: Espanso limits the regex matches to 30 characters ([Source](https://github.com/espanso/espanso/security)), so you must limit your query.

packages/bib-generator/1.0.0/_manifest.yml

@@ -0,0 +1,7 @@
+name: "bib-generator"
+title: "Bib Generator"
+homepage: https://github.com/Xeophon/hub/tree/main/packages/bib-generator
+description: Generate a BibTex entry by searching through DBLP
+version: 1.0.0
+author: Xeophon
+tags: ["bib", "academia", "research"]

packages/bib-generator/1.0.0/bib_generator.py

@@ -0,0 +1,55 @@
+import sys
+import urllib.request
+import urllib.parse
+import re
+
+
+def get_bib_info(bib):
+    """Extracts author, title, and year from a given bib entry."""
+    author = re.search(r"author\s*=\s*{(.*?)}", bib, re.DOTALL)
+    if author is None:
+        author = re.search(r"editor\s*=\s*{(.*?)}", bib, re.DOTALL)
+    author = author.group(1) if author is not None else "Unknown"
+    title = re.search(r"title\s*=\s*{(.*?)}", bib, re.DOTALL).group(1)
+    year = re.search(r"year\s*=\s*{(.*?)}", bib).group(1)
+    return author, title, year
+
+
+def format_bib_key(author, title, year):
+    """Formats the bib key using author's last name, first three words of the title, and the year."""
+    author_lastname = author.split(" and")[0].split()[-1].lower()
+    first_three_words = "".join(
+        re.sub(r"[^a-zA-Z]", "", word) for word in title.split()[:3]
+    )
+    key = f"{author_lastname}{first_three_words}{year}"
+    return key
+
+
+def extract_url(selected_entry):
+    """Extracts the URL part from the selected entry in the espanso form"""
+    pattern = r"\(([^()]*\/.*\/[^()]*)\)$"
+    match = re.search(pattern, selected_entry)
+    if match:
+        return match.group(1)
+    else:
+        return None
+
+
+def get_bib(selected_entry):
+    """Gets the bib from the selected entry in the espanso form"""
+    full_url = f"https://dblp.org/rec/{extract_url(selected_entry)}.bib"
+    with urllib.request.urlopen(full_url) as bib_response:
+        bib = bib_response.read().decode()
+        if "not found" not in bib:
+            author, title, year = get_bib_info(bib)
+            key = format_bib_key(author, title, year)
+            output = re.sub(r"\{DBLP:.*?\,", "{" + key + ",", bib)
+        else:
+            output = "Not found in DBLP"
+    print(output)
+
+
+try:
+    get_bib(sys.argv[1])
+except Exception as e:
+    output = str(e)

packages/bib-generator/1.0.0/dblp_multi_search.py

@@ -0,0 +1,29 @@
+import urllib.request
+import urllib.parse
+import json
+import sys
+
+
+def search(query):
+    """Searches for a publication and retrieves the top 5 matches."""
+    options = {"q": query, "format": "json", "h": 5}
+    output = ""
+    url = f"https://dblp.org/search/publ/api?{urllib.parse.urlencode(options)}"
+    with urllib.request.urlopen(url) as response:
+        data = json.loads(response.read().decode())
+        hit = data.get("result", {}).get("hits", {}).get("hit", [])
+
+        if hit:
+            for item in hit:
+                info = item.get("info", {})
+                title = info.get("title", "No title found")
+                output += f"{title} ({info.get('url').replace("https://dblp.org/rec/", "")})\n"
+        else:
+            print("No results found in DBLP")
+    print(output.strip())
+
+
+try:
+    search(sys.argv[1])
+except Exception as e:
+    output = str(e)

packages/bib-generator/1.0.0/package.yml

@@ -0,0 +1,28 @@
+matches:
+  - regex: ":cite:(?P<citation>.*?):"
+    replace: "{{final}}"
+    vars:
+      - name: dblp_search_results
+        type: script
+        params:
+          args:
+            - python
+            - "%CONFIG%/match/packages/bib-generator/dblp_multi_search.py"
+            - "{{citation}}"
+      - name: form1
+        type: form
+        params:
+          layout: |
+            Select bib entry:
+            [[entries]]
+          fields:
+            entries:
+              type: list
+              values: "{{dblp_search_results}}"
+      - name: final
+        type: script
+        params:
+          args:
+            - python
+            - "%CONFIG%/match/packages/bib-generator/bib_generator.py"
+            - "{{form1.entries}}"