Skip to content

Commit f2ee552

Browse files
AzzaareAzzaare
committed
Add flexible rules check. Tag new version. Enforce Julia ^1.6 compat
1 parent 12c10c6 commit f2ee552

File tree

3 files changed

+50
-31
lines changed

3 files changed

+50
-31
lines changed

‎Project.toml

+3-3
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
name = "BibParser"
22
uuid = "13533e5b-e1c2-4e57-8cef-cac5e52f6474"
33
authors = ["azzaare <[email protected]>"]
4-
version = "0.2.0"
4+
version = "0.2.1"
55

66
[deps]
77
BibInternal = "2027ae74-3657-4b95-ae00-e2f7d55c3e64"
@@ -11,8 +11,8 @@ JSONSchema = "7d188eb4-7ad8-530c-ae41-71a32a6d4692"
1111
YAML = "ddb6d928-2868-570f-bddf-ab3f9cf99eb6"
1212

1313
[compat]
14-
BibInternal = "0.3"
14+
BibInternal = "0.3.2"
1515
DataStructures = "0.18"
1616
JSONSchema = "1"
1717
YAML = "0.4"
18-
julia = "1.4"
18+
julia = "1.6"

‎src/BibParser.jl

+11-9
Original file line numberDiff line numberDiff line change
@@ -7,32 +7,34 @@ include("utils.jl")
77

88
# BibTeX module
99
include("bibtex.jl")
10-
import .BibTeX
10+
using .BibTeX: BibTeX
1111

1212
# CSL-JSON module
1313
include("csl.jl")
14-
import .CSL
14+
using .CSL: CSL
1515

1616
# CFF module
1717
include("cff.jl")
18-
import .CFF
18+
using .CFF: CFF
1919

2020
"""
2121
parse_file(path::String; parser::Symbol = :BibTeX)
2222
Parse a bibliography file. Default to BibTeX format. Other options available: CFF (CSL-JSON coming soon).
23+
For bibliography formats with formatting rules (such as `:BibTeX`), the `check` keyword argument can be set to `:none` (or `nothing`), `:warn`, or `:error`.
2324
"""
24-
parse_file(path, ::Val{:BibTeX}) = BibTeX.parse_file(path)
25-
parse_file(path, ::Val{:CFF}) = CFF.parse_file(path)
26-
27-
parse_file(path, parser = :BibTeX) = parse_file(path, Val(parser))
25+
parse_file(path, ::Val{:BibTeX}; check) = BibTeX.parse_file(path; check)
26+
parse_file(path, ::Val{:CFF}; check) = CFF.parse_file(path)
2827

28+
parse_file(path, parser=:BibTeX; check=:error) = parse_file(path, Val(parser); check)
2929

3030
"""
3131
parse_entry(entry::String; parser::Symbol = :BibTeX)
3232
Parse a string entry. Default to BibTeX format. No other options available yet (CSL-JSON coming soon).
33+
34+
For bibliography formats with formatting rules (such as `:BibTeX`), the `check` keyword argument can be set to `:none` (or `nothing`), `:warn`, or `:error`.
3335
"""
34-
function parse_entry(entry; parser=:BibTeX)
35-
return parser == :BibTeX && return BibTeX.parse_string(entry)
36+
function parse_entry(entry; parser=:BibTeX, check = :error)
37+
return parser == :BibTeX && return BibTeX.parse_string(entry; check)
3638
end
3739

3840
end # module

‎src/bibtex.jl

+36-19
Original file line numberDiff line numberDiff line change
@@ -216,6 +216,7 @@ A structure allowing to parse a BibTeX formatted string one character at a time.
216216
- `pos_start::Position`: pointer to the raw/col start position
217217
- `pos_end::Position`: pointer to the raw/col end position
218218
- `storage::Storage`: temporary storage of the content of an entry being parsed
219+
- `rules_checker::Bool`: indicate which level `:error`, `:warn`, or `:none` should be raised when an entry violates BibTeX rules
219220
- `task::Symbol`: describe which part of the BibTeX gramma is being parsed
220221
"""
221222
mutable struct Parser
@@ -226,6 +227,7 @@ mutable struct Parser
226227
input::Vector{Char}
227228
pos_start::Position
228229
pos_end::Position
230+
rules_checker::Symbol
229231
storage::Storage
230232
task::Symbol
231233

@@ -237,11 +239,21 @@ mutable struct Parser
237239
field=Field(),
238240
pos_start=Position(1, 1),
239241
pos_end=Position(1, 0),
242+
rules_checker=:error,
240243
storage=Storage(),
241244
task=:free,
242245
)
243246
return new(
244-
acc, content, errors, field, collect(input), pos_start, pos_end, storage, task
247+
acc,
248+
content,
249+
errors,
250+
field,
251+
collect(input),
252+
pos_start,
253+
pos_end,
254+
rules_checker,
255+
storage,
256+
task,
245257
)
246258
end
247259
end
@@ -326,10 +338,10 @@ function inc!(parser, char, dumped)
326338
end
327339
end
328340

329-
is_dumped(parser, char, ::Val{:free}) = char == '@'
341+
is_dumped(::Parser, char, ::Val{:free}) = char == '@'
330342
dump!(parser, char, ::Val) = char == '@' && (parser.task = :entry)
331343

332-
is_dumped(parser, char, ::Val{:entry}) = occursin(r"[@{\(\n]", char)
344+
is_dumped(::Parser, char, ::Val{:entry}) = occursin(r"[@{\(\n]", char)
333345
function dump!(parser, char, ::Val{:entry})
334346
if char == '\n'
335347
parser.task = :free
@@ -354,7 +366,7 @@ function dump!(parser, char, ::Val{:entry})
354366
end
355367
end
356368

357-
is_dumped(parser, char, ::Val{:key}) = char ∈ ['@', ',']
369+
is_dumped(::Parser, char, ::Val{:key}) = char ∈ ['@', ',']
358370
function dump!(parser, char, ::Val{:key})
359371
if char == '@'
360372
parser.task = :entry
@@ -378,7 +390,7 @@ function dump!(parser, char, ::Val{:key})
378390
end
379391
end
380392

381-
is_dumped(parser, char, ::Val{:field_name}) = char ∈ ['=', '@']
393+
is_dumped(::Parser, char, ::Val{:field_name}) = char ∈ ['=', '@']
382394
function dump!(parser, char, ::Val{:field_name})
383395
if char == '@'
384396
parser.task = :entry
@@ -405,7 +417,7 @@ function dump!(parser, char, ::Val{:field_name})
405417
end
406418
end
407419

408-
is_dumped(parser, char, ::Val{:field_in}) = occursin(r"[0-9@a-zA-Z\"{]", char)
420+
is_dumped(::Parser, char, ::Val{:field_in}) = occursin(r"[0-9@a-zA-Z\"{]", char)
409421
function dump!(parser, char, ::Val{:field_in})
410422
if char == '@'
411423
parser.task = :entry
@@ -456,7 +468,7 @@ function dump!(parser, char, ::Val{:field_inbrace})
456468
end
457469
end
458470

459-
is_dumped(parser, char, ::Val{:field_inquote}) = char ∈ ['{', '"', '}']
471+
is_dumped(::Parser, char, ::Val{:field_inquote}) = char ∈ ['{', '"', '}']
460472
function dump!(parser, char, ::Val{:field_inquote})
461473
if char == '"' && parser.field.braces == 0
462474
parser.field.value *= get_acc(parser; from=2)
@@ -489,18 +501,19 @@ function dump!(parser, char, ::Val{:field_outquote})
489501
parser.task = :field_next
490502
elseif char == rev(parser.storage.delim)
491503
entry = make_entry(parser.storage)
504+
check = parser.rules_checker
492505
push!(
493506
parser.content.entries,
494507
parser.storage.key =>
495-
BibInternal.make_bibtex_entry(parser.storage.key, entry),
508+
BibInternal.make_bibtex_entry(parser.storage.key, entry; check),
496509
)
497510
parser.storage = Storage()
498511
parser.task = :free
499512
end
500513
end
501514
end
502515

503-
is_dumped(parser, char, ::Val{:field_concat}) = occursin(r"[a-zA-Z\"@]", char)
516+
is_dumped(::Parser, char, ::Val{:field_concat}) = occursin(r"[a-zA-Z\"@]", char)
504517
function dump!(parser, char, ::Val{:field_concat})
505518
if char == '@'
506519
parser.task = :entry
@@ -540,10 +553,11 @@ function dump!(parser, char, ::Val{:field_var})
540553
parser.task = :field_next
541554
elseif char == rev(parser.storage.delim)
542555
entry = make_entry(parser.storage)
556+
check = parser.rules_checker
543557
push!(
544558
parser.content.entries,
545559
parser.storage.key =>
546-
BibInternal.make_bibtex_entry(parser.storage.key, entry),
560+
BibInternal.make_bibtex_entry(parser.storage.key, entry; check),
547561
)
548562
parser.storage = Storage()
549563
parser.task = :free
@@ -579,10 +593,11 @@ function dump!(parser, char, ::Val{:field_number})
579593
parser.task = :field_next
580594
elseif char == rev(parser.storage.delim)
581595
entry = make_entry(parser.storage)
596+
check = parser.rules_checker
582597
push!(
583598
parser.content.entries,
584599
parser.storage.key =>
585-
BibInternal.make_bibtex_entry(parser.storage.key, entry),
600+
BibInternal.make_bibtex_entry(parser.storage.key, entry; check),
586601
)
587602
parser.task = :free
588603
end
@@ -608,9 +623,10 @@ function dump!(parser, char, ::Val{:field_out})
608623
parser.task = :field_next
609624
elseif char == rev(parser.storage.delim)
610625
entry = make_entry(parser.storage)
626+
check = parser.rules_checker
611627
push!(
612628
parser.content.entries,
613-
parser.storage.key => BibInternal.make_bibtex_entry(parser.storage.key, entry),
629+
parser.storage.key => BibInternal.make_bibtex_entry(parser.storage.key, entry; check),
614630
)
615631
parser.storage = Storage()
616632
parser.task = :free
@@ -641,16 +657,17 @@ function dump!(parser, char, ::Val{:field_next})
641657
end
642658
elseif char == rev(parser.storage.delim)
643659
entry = make_entry(parser.storage)
660+
check = parser.rules_checker
644661
push!(
645662
parser.content.entries,
646-
parser.storage.key => BibInternal.make_bibtex_entry(parser.storage.key, entry),
663+
parser.storage.key => BibInternal.make_bibtex_entry(parser.storage.key, entry; check),
647664
)
648665
parser.storage = Storage()
649666
parser.task = :free
650667
end
651668
end
652669

653-
is_dumped(parser, char, ::Val{:string}) = char ∈ ['=', '@']
670+
is_dumped(::Parser, char, ::Val{:string}) = char ∈ ['=', '@']
654671
function dump!(parser, char, ::Val{:string})
655672
if char == '@'
656673
parser.task = :entry
@@ -677,7 +694,7 @@ function dump!(parser, char, ::Val{:string})
677694
end
678695
end
679696

680-
is_dumped(parser, char, ::Val{:string_inquote}) = char ∈ ['"', '@']
697+
is_dumped(::Parser, char, ::Val{:string_inquote}) = char ∈ ['"', '@']
681698
function dump!(parser, char, ::Val{:string_inquote})
682699
if char == '@'
683700
parser.task = :entry
@@ -690,7 +707,7 @@ function dump!(parser, char, ::Val{:string_inquote})
690707
end
691708
end
692709

693-
is_dumped(parser, char, ::Val{:string_value}) = char ∈ ['"']
710+
is_dumped(::Parser, char, ::Val{:string_value}) = char ∈ ['"']
694711
function dump!(parser, char, ::Val{:string_value})
695712
if char == '"'
696713
parser.field.value = get_acc(parser; from=2)
@@ -747,8 +764,8 @@ end
747764
748765
Parse a BibTeX string of entries. Raise a detailed warning for each invalid entry.
749766
"""
750-
function parse_string(str)
751-
parser = Parser(str)
767+
function parse_string(str; check=:error)
768+
parser = Parser(str; rules_checker=check)
752769
foreach(char -> parse!(parser, char), parser.input)
753770
foreach(error -> warn(error), parser.errors)
754771
return get_entries(parser)
@@ -759,6 +776,6 @@ end
759776
760777
Parse a BibTeX file located at `path`. Raise a detailed warning for each invalid entry.
761778
"""
762-
parse_file(path) = parse_string(read(path, String))
779+
parse_file(path; check=:error) = parse_string(read(path, String); check)
763780

764781
end # module

0 commit comments

Comments
 (0)