bring formatting to utils

mkdocs_bibtex/utils.py

@@ -1,65 +1,112 @@
 import re
 import tempfile
+from collections import OrderedDict
 from itertools import groupby
 from pathlib import Path
 
 import pypandoc
+from pybtex.backends.markdown import Backend as MarkdownBackend
 from pybtex.database import BibliographyData
+from pybtex.style.formatting.plain import Style as PlainStyle
 
 
-def to_markdown_pandoc(entry, csl_path):
+def format_simple(entries):
+    """
+    Format the entries using a simple built in style
+
+    Args:
+        entries (dict): dictionary of entries
+
+    Returns:
+        references (dict): dictionary of citation texts
+    """
+    style = PlainStyle()
+    backend = MarkdownBackend()
+    citations = OrderedDict()
+    for key, entry in entries.items():
+        formatted_entry = style.format_entry("", entry)
+        entry_text = formatted_entry.text.render(backend)
+        entry_text = entry_text.replace("\n", " ")
+        # Local reference list for this file
+        citations[key] = entry_text
+    return citations
+
+
+def format_pandoc(entries, csl_path):
+    """
+    Format the entries using pandoc
+
+    Args:
+        entries (dict): dictionary of entries
+        csl_path (str): path to formatting CSL Fle
+    Returns:
+        references (dict): dictionary of citation texts
+    """
+    pandoc_version = tuple(int(ver) for ver in pypandoc.get_pandoc_version().split("."))
+    citations = OrderedDict()
+    for key, entry in entries.items():
+        bibtex_string = BibliographyData(entries={entry.key: entry}).to_string("bibtex")
+        if pandoc_version >= (2, 11):
+            citations[key] = _convert_pandoc_new(bibtex_string, csl_path)
+        else:
+            citations[key] = _convert_pandoc_legacy(bibtex_string, csl_path)
+
+    return citations
+
+
+def _convert_pandoc_new(bibtex_string, csl_path):
     """
     Converts the PyBtex entry into formatted markdown citation text
+    using pandoc version 2.11 or newer
     """
-    bibtex_string = BibliographyData(entries={entry.key: entry}).to_string("bibtex")
-    if tuple(int(ver) for ver in pypandoc.get_pandoc_version().split(".")) >= (
-        2,
-        11,
-    ):
-        markdown = pypandoc.convert_text(
-            source=bibtex_string,
-            to="markdown-citations",
-            format="bibtex",
-            extra_args=[
-                "--citeproc",
-                "--csl",
-                csl_path,
-            ],
-        )
+    markdown = pypandoc.convert_text(
+        source=bibtex_string,
+        to="markdown-citations",
+        format="bibtex",
+        extra_args=[
+            "--citeproc",
+            "--csl",
+            csl_path,
+        ],
+    )
+
+    # This should cut off the pandoc preamble and ending triple colons
+    markdown = " ".join(markdown.split("\n")[2:-2])
+
+    citation_regex = re.compile(r"\{\.csl-left-margin\}\[(.*)\]\{\.csl-right-inline\}")
+    try:
+
+        citation = citation_regex.findall(markdown)[0]
+    except IndexError:
+        citation = markdown
+    return citation
 
-        # This should cut off the pandoc preamble and ending triple colons
-        markdown = " ".join(markdown.split("\n")[2:-2])
 
-        citation_regex = re.compile(
-            r"\{\.csl-left-margin\}\[(.*)\]\{\.csl-right-inline\}"
-        )
-        try:
-
-            citation = citation_regex.findall(markdown)[0]
-        except IndexError:
-            citation = markdown
-    else:
-        # Older citeproc-filter version of pandoc
-        with tempfile.TemporaryDirectory() as tmpdir:
-            bib_path = Path(tmpdir).joinpath("temp.bib")
-            with open(bib_path, "w") as bibfile:
-                bibfile.write(bibtex_string)
-            citation_text = """
+def _convert_pandoc_legacy(bibtex_string, csl_file):
+    """
+    Converts the PyBtex entry into formatted markdown citation text
+    using pandoc version older than 2.11
+    """
+    with tempfile.TemporaryDirectory() as tmpdir:
+        bib_path = Path(tmpdir).joinpath("temp.bib")
+        with open(bib_path, "w") as bibfile:
+            bibfile.write(bibtex_string)
+        citation_text = """
 ---
 nocite: '@*'
 ---
 """
 
-            markdown = pypandoc.convert_text(
-                source=citation_text,
-                to="markdown_strict",
-                format="md",
-                extra_args=["--csl", csl_path, "--bibliography", bib_path],
-                filters=["pandoc-citeproc"],
-            )
+        markdown = pypandoc.convert_text(
+            source=citation_text,
+            to="markdown_strict",
+            format="md",
+            extra_args=["--csl", csl_path, "--bibliography", bib_path],
+            filters=["pandoc-citeproc"],
+        )
 
-        citation_regex = re.compile(r"(1\.)?(.*)")
-        citation = citation_regex.findall(markdown.replace("\n", " "))[0]
+    citation_regex = re.compile(r"(1\.)?(.*)")
+    citation = citation_regex.findall(markdown.replace("\n", " "))[0]
     return citation
 
 

test_files/test_plugin.py

@@ -2,9 +2,14 @@
 
 import pytest
 
-from mkdocs_bibtex.plugin import BibTexPlugin
-from mkdocs_bibtex.utils import (find_cite_keys, format_bibliography,
-                                 insert_citation_keys)
+from mkdocs_bibtex.plugin import BibTexPlugin, parse_file
+from mkdocs_bibtex.utils import (
+    find_cite_keys,
+    format_bibliography,
+    insert_citation_keys,
+    format_simple,
+    format_pandoc,
+)
 
 module_dir = os.path.dirname(os.path.abspath(__file__))
 test_files_dir = os.path.abspath(os.path.join(module_dir, "..", "test_files"))
@@ -21,6 +26,12 @@ def plugin():
     return plugin
 
 
+@pytest.fixture
+def entries():
+    bibdata = parse_file(os.path.join(test_files_dir, "test.bib"))
+    return bibdata.entries
+
+
 def test_bibtex_loading_bibfile(plugin):
     assert len(plugin.bib_data.entries) == 3
 
@@ -180,4 +191,38 @@ def test_format_bibliography():
     assert (
         "[^2]: First Author and Second Author\\. Test Title \\(TT\\)\\. *Testing Journal \\(TJ\\)*, 2019"
         in bib
-    )
+    )
+
+
+def test_format_simple(entries):
+    citations = format_simple(entries)
+
+    assert all(k in citations for k in entries)
+    assert all(entry != citations[k] for k, entry in entries.items())
+
+    print(citations)
+    assert (
+        citations["test"]
+        == "First Author and Second Author\\. Test title\\. *Testing Journal*, 2019\\."
+    )
+    assert (
+        citations["test2"]
+        == "First Author and Second Author\\. Test Title \\(TT\\)\\. *Testing Journal \\(TJ\\)*, 2019\\."
+    )
+
+
+def test_format_pandoc(entries):
+    citations = format_pandoc(entries, os.path.join(test_files_dir, "nature.csl"))
+
+    assert all(k in citations for k in entries)
+    assert all(entry != citations[k] for k, entry in entries.items())
+
+    print(citations)
+    assert (
+        citations["test"]
+        == "Author, F. & Author, S. Test title. *Testing Journal* **1**, (2019)."
+    )
+    assert (
+        citations["test2"]
+        == "Author, F. & Author, S. Test Title (TT). *Testing Journal (TJ)* **1**, (2019)."
+    )