Define citekey aliases with link reference syntax

Inspired by https://github.com/phiresky/pandoc-url2cite

Refs phiresky/pandoc-url2cite#1

manubot/pandoc/cite.py

@@ -21,13 +21,12 @@
 http://scorreia.com/software/panflute/code.html#panflute.elements.Citation
 """
 import argparse
+import logging
 import sys
 
 import panflute as pf
 
-from manubot.cite.citekey import (
-    is_valid_citekey,
-)
+from manubot.cite.citekey import is_valid_citekey
 from manubot.process.util import (
     get_citekeys_df,
     generate_csl_items,
@@ -65,7 +64,7 @@ def parse_args():
     return args
 
 
-def _get_citation_string_action(elem, doc):
+def _get_citekeys_action(elem, doc):
     """
     Panflute action to extract citationId from all Citations in the AST.
     """
@@ -89,6 +88,39 @@ def _citation_to_id_action(elem, doc):
     return None
 
 
+def _get_reference_link_citekey_aliases(elem, doc):
+    """
+    Based on TypeScript implementation by phiresky at
+    https://github.com/phiresky/pandoc-url2cite/blob/b28374a9a037a5ce1747b8567160d8dffd64177e/index.ts#L118-L152
+
+    Uses markdown's link reference syntax to define citekey aliases (tags)
+    https://spec.commonmark.org/0.29/#link-reference-definitions
+    """
+    if type(elem) != pf.Para:
+        return
+    while (
+        len(elem.content) >= 3
+        and type(elem.content[0]) == pf.Cite
+        and len(elem.content[0].citations) == 1  # differs from pandoc-url2cite
+        and type(elem.content[1]) == pf.Str
+        and elem.content[1].text == ":"
+    ):
+        space_index = 3 if type(elem.content[2]) == pf.Space else 2
+        destination = elem.content[space_index]
+        if type(destination) == pf.Str:
+            # paragraph starts with [@something]: something
+            # save info to citekeys and remove from paragraph
+            citekey = elem.content[0].citations[0].id  # differs from pandoc-url2cite
+            citekey_aliases = global_variables["citekey_aliases"]
+            if citekey in citekey_aliases:
+                logging.warning(f"duplicate citekey {citekey}")
+            citekey_aliases[citekey] = destination.text
+            # found citation, add it to citekeys and remove it from document
+            elem.content = elem.content[space_index + 1 :]
+        if len(elem.content) > 0 and type(elem.content[0]) == pf.SoftBreak:
+            elem.content.pop(0)
+
+
 def process_citations(doc):
     """
     Apply citation-by-identifier to a Python object representation of
@@ -100,8 +132,13 @@ def process_citations(doc):
     - manubot-requests-cache-path
     - manubot-clear-requests-cache
     """
+    global_variables["citekey_aliases"] = doc.get_metadata(
+        "citekey-aliases", default={}, builtin=True
+    )
+    manuscript_citekeys = set(global_variables["manuscript_citekeys"])
 
-    doc.walk(_get_citation_string_action)
+    doc.walk(_get_reference_link_citekey_aliases)
+    doc.walk(_get_citekeys_action)
     manuscript_citekeys = set(global_variables["manuscript_citekeys"])
     manuscript_citekeys = sorted(
         filter(
@@ -112,8 +149,9 @@ def process_citations(doc):
         )
     )
     global_variables["manuscript_citekeys"] = manuscript_citekeys
-    tag_to_string = doc.get_metadata("citekey-aliases", default={}, builtin=True)
-    citekeys_df = get_citekeys_df(manuscript_citekeys, tag_to_string)
+    citekeys_df = get_citekeys_df(
+        manuscript_citekeys, global_variables["citekey_aliases"]
+    )
     global_variables["citekeys_df"] = citekeys_df
     global_variables["citekey_shortener"] = dict(
         zip((citekeys_df["manuscript_citekey"]), citekeys_df["short_citekey"])

manubot/pandoc/tests/input-with-cites.md

@@ -8,6 +8,13 @@ This is a sentence with many citations [@pmid:20170387; @doi:10.7717/peerj.705].
 Citations in code blocks should not be modified [@pmid:20170387].
 ```
 
+Defining citekeys with the link reference syntax [@tag:issue; @tag:bad-doi; @tag:bad-url].
+
+[@tag:issue]: url:https://github.com/manubot/manubot/pull/189
+[@tag:bad-doi]: doi:10.1016/S0022-2836(05)80360-2
+
+[@tag:bad-url]: url:https://openreview.net/forum?id=HkwoSDPgg
+
 ## References
 
 ::: {#refs}

manubot/pandoc/tests/output-with-cites.txt

@@ -6,6 +6,8 @@ This is a sentence with many citations [1,2].
 
     Citations in code blocks should not be modified [@pmid:20170387].
 
+Defining citekeys with the link reference syntax [3,4,5].
+
 References
 
 1. Orthodontic treatment with tooth transplantation for patients with cleft lip and palate.
@@ -17,3 +19,14 @@ DOI: 10.1597/08-134 · PMID: 20170387
 Kamen P. Simeonov, Daniel S. Himmelstein
 PeerJ (2015-01-13) https://doi.org/98p
 DOI: 10.7717/peerj.705 · PMID: 25648772 · PMCID: PMC4304851
+
+3. CSL_Item: add date IO functionality by dhimmel · Pull Request #189 · manubot/manubotGitHub https://github.com/manubot/manubot/pull/189
+
+4. Basic local alignment search tool
+Stephen F. Altschul, Warren Gish, Webb Miller, Eugene W. Myers, David J. Lipman
+Journal of Molecular Biology (1990-10) https://doi.org/cnsjsz
+DOI: 10.1016/s0022-2836(05)80360-2
+
+5. Semi-supervised Knowledge Transfer for Deep Learning from Private Training Data
+Nicolas Papernot, Martí, N Abadi, Ú, Lfar Erlingsson, Ian Goodfellow, Kunal Talwar
+(2016-11-02) https://openreview.net/forum?id=HkwoSDPgg