Parse biblatex fields date, journaltitle and location in bibtex readers

NEWS.md

@@ -4,6 +4,9 @@
 
 - fixed issue GeoBosh/Rdpack#40.
 
+- bibtex readers now parse the biblatex fields `date`, `journaltitle` and
+  `location`.
+
 
 # rbibutils 2.4.1
 

inst/bib/00README.org

@@ -1,3 +1,9 @@
+# 2026-09-16
+
+  =biblatex_aliases.bib= tests biblatex fields 'date', 'journaltitle' and 'location'
+  in the bibtex readers.
+
+
 # 2025-11-04
 
   add ~@string{j-GUTENBERG = Gutenberg journal}~ at the beginning of Cousquer*

inst/bib/biblatex_aliases.bib

@@ -0,0 +1,25 @@
+@article{full,
+  author = {Smith, John}, title = {Full date},
+  JournalTitle = {Some Journal}, Date = {2020-02-02}, volume = {3}
+}
+@article{yearmonth,
+  author = {Jones, Ann}, title = {Year and month},
+  journaltitle = {J}, date = "2019-07"
+}
+@article{precedence,
+  author = {Doe, Jane}, title = {Biblatex fields take precedence},
+  journal = {Old Journal}, journaltitle = {New Journal},
+  year = {1999}, month = {dec}, day = {31}, date = {2001~/2002-03}
+}
+@book{location,
+  author = {Brown, Bob}, title = {Location as address}, publisher = {P},
+  location = {Durham}, date = {2021}
+}
+@book{address,
+  author = {Brown, Bob}, title = {Address wins, unparsable date ignored}, publisher = {P},
+  address = {London}, location = {Durham}, year = 2005, date = {circa 2004}
+}
+@patent{patent,
+  author = {Green, Gill}, title = {A patent}, number = {1},
+  location = {countryfr}, date = {2010-05}
+}

man/readBib.Rd

@@ -93,6 +93,12 @@ charToBib(text, informat, \dots)
   non-standard fields in the bib entries are preserved in the bibentry
   object.
 
+  Following pandoc and biber, \code{date} takes precedence over \code{year},
+  \code{month} and \code{day}, and \code{journaltitle} overrides
+  \code{journal}. \code{location} is used for \code{address} if the
+  latter is missing (except for patents). These biblatex fields are kept
+  when \code{direct = TRUE}.
+
   Argument \code{texChars}, currently implemented only for the case
   \code{direct = TRUE}, gives some control over the processing of TeX
   sequences representing characters (such as accented Latin characters):

src/common_bt_btd.c

@@ -10,6 +10,9 @@
  *
  */
 
+#include <ctype.h>
+#include <string.h>
+
 #include "str.h"
 #include "fields.h"
 #include "slist.h"
@@ -21,6 +24,128 @@
 extern slist find;
 extern slist replace;
 
+/* 2026-09-16: biblatex aliases
+ *
+ * Biblatex (and hence e.g. Zotero's Better BibTeX export) uses fields 'date' and
+ * 'journaltitle' instead of bibtex's 'year'/'month'/'day' and 'journal', and 'location'
+ * as an alias of 'address'.  Without 'year' and 'journal', R's bibentry() refuses to
+ * create entries of type 'Article', etc.
+ *
+ * Mirror pandoc (and biber): if present, 'date' and 'journaltitle' take precedence over
+ * 'year'/'month'/'day' and 'journal'; 'location' is used only if 'address' is missing
+ * (and not for patents).
+ * The biblatex fields themselves are kept.
+ */
+
+static const char *biblatex_month_names[12] = {
+  "January", "February", "March", "April", "May", "June",
+  "July", "August", "September", "October", "November", "December"
+};
+
+/* Parse the start of a biblatex/EDTF date: [-]YYYY[-MM[-DD]], optionally followed by
+ * an end of range ('/...'), a time ('T...') or qualifiers ('?', '~', '%').
+ * Months 21-24 (seasons) are ignored.
+ * Returns 1 on success, 0 if 'date' is not recognised (then it is ignored).
+ */
+static int
+parse_biblatex_date( const char *p, str *year, str *month, str *day )
+{
+	int m = 0;
+	const char *q;
+
+	str_empty( year ); str_empty( month ); str_empty( day );
+
+	while ( isspace( (unsigned char) *p ) ) p++;
+
+	q = p;
+	if ( *q == '-' ) q++;
+	if ( !( isdigit( (unsigned char) q[0] ) && isdigit( (unsigned char) q[1] ) &&
+		isdigit( (unsigned char) q[2] ) && isdigit( (unsigned char) q[3] ) ) )
+		return 0;
+	q += 4;
+	str_segcpy( year, (char *) p, (char *) q );
+
+	if ( q[0] == '-' && isdigit( (unsigned char) q[1] ) && isdigit( (unsigned char) q[2] ) ) {
+		m = 10 * ( q[1] - '0' ) + ( q[2] - '0' );
+		q += 3;
+		if ( m >= 1 && m <= 12 ) {
+			str_strcpyc( month, biblatex_month_names[m - 1] );
+			if ( q[0] == '-' && isdigit( (unsigned char) q[1] ) &&
+			     isdigit( (unsigned char) q[2] ) ) {
+				str_segcpy( day, (char *) q + 1, (char *) q + 3 );
+				q += 3;
+			}
+		} else if ( m < 21 || m > 24 ) {
+			return 0;
+		}
+	}
+
+	while ( *q == '?' || *q == '~' || *q == '%' ) q++;
+	while ( isspace( (unsigned char) *q ) ) q++;
+	if ( *q != '\0' && *q != '/' && *q != 'T' ) return 0;
+
+	return !str_memerr( year ) && !str_memerr( month ) && !str_memerr( day );
+}
+
+/* set field 'tag' to 'value', or remove all instances of 'tag' if 'value' is empty */
+static int
+biblatex_set_field( fields *bibin, const char *tag, str *value )
+{
+	int i;
+
+	for ( i = fields_num( bibin ) - 1; i >= 0; i-- ) {
+		if ( !fields_match_casetag( bibin, i, tag ) ) continue;
+		if ( fields_remove( bibin, i ) != FIELDS_OK ) return BIBL_ERR_MEMERR;
+	}
+	if ( str_has_value( value ) &&
+	     fields_add( bibin, tag, str_cstr( value ), LEVEL_MAIN ) != FIELDS_OK )
+		return BIBL_ERR_MEMERR;
+
+	return BIBL_OK;
+}
+
+static int
+process_biblatex_aliases( fields *bibin )
+{
+	int n, status = BIBL_OK;
+	str year, month, day, value;
+
+	strs_init( &year, &month, &day, &value, NULL );
+
+	n = fields_find( bibin, "date", LEVEL_MAIN );
+	if ( n != FIELDS_NOTFOUND &&
+	     parse_biblatex_date( fields_value( bibin, n, FIELDS_CHRP ), &year, &month, &day ) ) {
+		status = biblatex_set_field( bibin, "year", &year );
+		if ( status == BIBL_OK ) status = biblatex_set_field( bibin, "month", &month );
+		if ( status == BIBL_OK ) status = biblatex_set_field( bibin, "day", &day );
+		if ( status != BIBL_OK ) goto out;
+	}
+
+	n = fields_find( bibin, "journaltitle", LEVEL_MAIN );
+	if ( n != FIELDS_NOTFOUND ) {
+		str_strcpyc( &value, fields_value( bibin, n, FIELDS_CHRP ) );
+		status = biblatex_set_field( bibin, "journal", &value );
+		if ( status != BIBL_OK ) goto out;
+	}
+
+	/* for patents biblatex's 'location' is the jurisdiction, not an address */
+	n = fields_find( bibin, "INTERNAL_TYPE", LEVEL_MAIN );
+	if ( n != FIELDS_NOTFOUND && !strcasecmp( fields_value( bibin, n, FIELDS_CHRP ), "patent" ) )
+		goto out;
+
+	if ( fields_find( bibin, "address", LEVEL_MAIN ) == FIELDS_NOTFOUND ) {
+		n = fields_find( bibin, "location", LEVEL_MAIN );
+		if ( n != FIELDS_NOTFOUND &&
+		     fields_add( bibin, "address", fields_value( bibin, n, FIELDS_CHRP ),
+				 LEVEL_MAIN ) != FIELDS_OK )
+			status = BIBL_ERR_MEMERR;
+	}
+
+out:
+	strs_free( &year, &month, &day, &value, NULL );
+	return status;
+}
+
 /* process_ref()
  *
  */
@@ -54,6 +179,8 @@ process_ref( fields *bibin, const char *p, loc *currloc )
 		if ( fstatus!=FIELDS_OK ) { status = BIBL_ERR_MEMERR; goto out; }
 
 	}
+
+	status = process_biblatex_aliases( bibin ); // 2026-09-16
 out:
 	strs_free( &type, &id, &tag, &data, NULL );
 	return status;

tests/testthat/test-bib.R

@@ -278,3 +278,37 @@ test_that("bibRead works ok", {
     if(is.numeric(svnrev <- R.Version()$'svn rev')  &&  svnrev >= 84986)
         expect_known_value(accfn, "acc_fn.rds", FALSE)
 })
+
+test_that("biblatex fields date, journaltitle and location are understood", {
+    fn <- system.file("bib", "biblatex_aliases.bib", package = "rbibutils")
+    ## with direct = FALSE, the (pre-existing) conversion of type 'patent' to 'Misc'
+    ##   prints "Cannot identify TYPE in reference 6 patent"
+    for(direct in c(TRUE, FALSE)){
+        bib <- readBib(fn, direct = direct)
+        expect_setequal(names(bib), c("full", "yearmonth", "precedence",
+                                      "location", "address", "patent"))
+
+        expect_equal(bib[["full"]]$year, "2020")
+        expect_equal(bib[["full"]]$month, "February")
+        expect_equal(bib[["full"]]$day, "02")
+        expect_equal(bib[["full"]]$journal, "Some Journal")
+
+        expect_equal(bib[["yearmonth"]]$year, "2019")
+        expect_equal(bib[["yearmonth"]]$month, "July")
+        expect_null(bib[["yearmonth"]]$day)
+
+        ## date and journaltitle take precedence (as in pandoc and biber)
+        expect_equal(bib[["precedence"]]$year, "2001")
+        expect_null(bib[["precedence"]]$month)
+        expect_null(bib[["precedence"]]$day)
+        expect_equal(bib[["precedence"]]$journal, "New Journal")
+
+        expect_equal(bib[["location"]]$address, "Durham")
+        ## address takes precedence over location; unparsable date is ignored
+        expect_equal(bib[["address"]]$address, "London")
+        expect_equal(bib[["address"]]$year, "2005")
+
+        expect_null(bib[["patent"]]$address)
+        expect_equal(bib[["patent"]]$year, "2010")
+    }
+})