diff options
Diffstat (limited to 'org/default_regex.org')
| -rw-r--r-- | org/default_regex.org | 116 |
1 files changed, 78 insertions, 38 deletions
diff --git a/org/default_regex.org b/org/default_regex.org index d9020d4..d78409e 100644 --- a/org/default_regex.org +++ b/org/default_regex.org @@ -7,6 +7,7 @@ #+COPYRIGHT: Copyright (C) 2015 - 2020 Ralph Amissah #+LANGUAGE: en #+STARTUP: content hideblocks hidestars noindent entitiespretty +#+OPTIONS: H:3 num:nil toc:t \n:nil @:t ::t |:t ^:nil _:nil -:t f:t *:t <:t #+PROPERTY: header-args :exports code #+PROPERTY: header-args+ :noweb yes #+PROPERTY: header-args+ :eval no @@ -37,15 +38,45 @@ http://dlang.org/phobos/std_regex.html module doc_reform.meta.rgx; static template spineRgxIn() { static struct RgxI { - <<meta_rgx>> - <<prgmkup_rgx>> + <<meta_rgx_0>> + <<meta_rgx_1>> + <<meta_rgx_2>> + <<meta_rgx_3>> + <<meta_rgx_4>> + <<meta_rgx_5>> + <<meta_rgx_6>> + <<meta_rgx_7>> + <<meta_rgx_8>> + <<meta_rgx_9>> + <<meta_rgx_10>> + <<meta_rgx_11>> + <<meta_rgx_12>> + <<meta_rgx_13>> + <<meta_rgx_14>> + <<meta_rgx_15>> + <<meta_rgx_16>> + <<meta_rgx_17>> + <<meta_rgx_18>> + <<meta_rgx_19>> + <<meta_rgx_20>> + <<meta_rgx_21>> + <<meta_rgx_22>> + <<prgmkup_rgx_0>> + <<prgmkup_rgx_1>> + <<prgmkup_rgx_2>> + <<prgmkup_rgx_3>> + <<prgmkup_rgx_4>> + <<prgmkup_rgx_5>> + <<prgmkup_rgx_6>> + <<prgmkup_rgx_7>> + <<prgmkup_rgx_8>> } } #+END_SRC ** misc :misc: -#+NAME: meta_rgx +#+NAME: meta_rgx_0 #+BEGIN_SRC d /+ misc +/ // static true_dollar = ctRegex!(`\$`, "gm"); @@ -84,7 +115,7 @@ static numeric_col = ctRegex!(`^[ 0-9,.%$£ ** comments :comment: -#+NAME: meta_rgx +#+NAME: meta_rgx_1 #+BEGIN_SRC d /+ comments +/ static comment = ctRegex!(`^%+ `); @@ -92,7 +123,7 @@ static comment = ctRegex!(`^%+ `); ** config -#+NAME: meta_rgx +#+NAME: meta_rgx_2 #+BEGIN_SRC d /+ header +/ #+END_SRC @@ -100,7 +131,7 @@ static comment = ctRegex!(`^%+ `); ** native headers *** native header :native:header: -#+NAME: meta_rgx +#+NAME: meta_rgx_3 #+BEGIN_SRC d /+ header +/ static variable_doc_title = ctRegex!(`@title`); @@ -112,7 +143,7 @@ static yaml_config = ctRegex!(`^[a-z]+\s*:\s* ** heading & paragraph operators :paragraph:operator: -#+NAME: meta_rgx +#+NAME: meta_rgx_4 #+BEGIN_SRC d /+ heading & paragraph operators +/ static heading_a = ctRegex!(`^:?[A][~] `, "m"); @@ -141,7 +172,7 @@ static para_inline_link_anchor = ctRegex!(`\*[~](?P<ancho ** blocked markup *** blocked markup curly & tic :block: -#+NAME: meta_rgx +#+NAME: meta_rgx_5 #+BEGIN_SRC d /+ blocked markup +/ static block_open = ctRegex!("^((code(?:[.][a-z][0-9a-z#+_]+)?|(?:poem|group|block|quote)(?:[.][a-z][0-9a-z_]+)?|table)(?:[(][ a-zA-Z0-9;:,]*[)])?[{][ ]*$)|^`{3} (code(?:[.][a-z][0-9a-z#+_]+)?|(?:poem|group|block|quote)(?:[.][a-z][0-9a-z_]+)?|table)(?:[(][ a-zA-Z0-9;:,]*[)])?|^[{]table[(](?:h;)?(?P<columns>(?:[ ,]+[0-9]+)+)[)][}]"); @@ -150,7 +181,7 @@ static block_poem_open = ctRegex!("^((poem(?:[(][ *** blocked markup tic :block:tic: -#+NAME: meta_rgx +#+NAME: meta_rgx_6 #+BEGIN_SRC d /+ blocked markup tics +/ static block_tic_code_open = ctRegex!("^`{3} code(?:[.](?P<syntax>[a-z][0-9a-z#+_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?"); @@ -164,7 +195,7 @@ static block_tic_close = ctRegex!("^(`{3})$","m") *** blocked markup curly :block:curly: -#+NAME: meta_rgx +#+NAME: meta_rgx_7 #+BEGIN_SRC d /+ blocked markup curly +/ static block_curly_code_open = ctRegex!(`^(?:code(?:[.](?P<syntax>[a-z][0-9a-z_]+))?(?:[(](?P<attrib>[ a-zA-Z0-9;:,]*)[)])?[{][ ]*$)`); @@ -185,14 +216,14 @@ static block_curly_table_special_markup = ctRegex!(`^[{]table[(](? *** block sub-matches :block: **** code -#+NAME: meta_rgx +#+NAME: meta_rgx_8 #+BEGIN_SRC d static code_numbering = ctRegex!(`(?P<number>\blinenumber\b|\bnumber\b|\blnr\b)`); #+END_SRC **** table -#+NAME: meta_rgx +#+NAME: meta_rgx_9 #+BEGIN_SRC d static table_head_instructions = ctRegex!(`(?:(?P<c_heading>h);)?(?:[ ]+c(?P<c_num>[0-9]):)?(?P<c_widths>(?:[, ]+[0-9]+[lr]?)+)`); static table_col_widths_and_alignment = ctRegex!(`(?P<width>[0-9]+)(?P<align>[lr]?)`); @@ -206,7 +237,7 @@ static table_col_separator_nl = ctRegex!(`[┊]$`, "mg") ** inline markup :inline:footnote: *** footnotes & endnotes -#+NAME: meta_rgx +#+NAME: meta_rgx_10 #+BEGIN_SRC d /+ inline markup footnotes endnotes +/ static inline_notes_curly_gen = ctRegex!(`~\{.+?\}~`, "m"); @@ -224,7 +255,7 @@ static note_ref = ctRegex!(`^\S+?noteref_( *** links/ urls :inline:footnote: -#+NAME: meta_rgx +#+NAME: meta_rgx_11 #+BEGIN_SRC d static webserv_url_doc_root = ctRegex!(`(?P<url>(?P<domain>https?:\/\/[^ /]+)\/(?P<path>\S*))`, "mg"); static smid_inline_url_generic = ctRegex!(`(?:^|[}(\[ ])(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_#]`, "mg"); @@ -237,7 +268,7 @@ static smid_inline_link_endnote_url_helper = ctRegex!(`\{~\^\s+(?P<co *** images :images: -#+NAME: meta_rgx +#+NAME: meta_rgx_12 #+BEGIN_SRC d static image = ctRegex!(`([a-zA-Z0-9._-]+?\.(?:png|gif|jpg))`, "mg"); static smid_image = ctRegex!(`(?P<pre>(?:^|[ ]|[^\S]?)[{┥](?:~\^\s+|\s*))(?P<image>[a-zA-Z0-9._-]+?\.(?:png|gif|jpg))(?P<post>(?:.*?)\s*[}┝](?:image|┤.*?├|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[;:!,?.]?([ )\]]|$)))`, "mg"); @@ -250,7 +281,7 @@ static smid_image_delimit = ctRegex!(`(?P<pre>^|[ ] *** inline markup book index :inline:bookindex: -#+NAME: meta_rgx +#+NAME: meta_rgx_13 #+BEGIN_SRC d /+ inline markup book index +/ static book_index_item = ctRegex!(`^=\{\s*(?P<bookindex>.+?)\}$`, "m"); @@ -261,7 +292,7 @@ static book_index_item_close = ctRegex!(`^(.*?)\}$`, "m ** switch *** switch off auto-heading number -#+NAME: meta_rgx +#+NAME: meta_rgx_14 #+BEGIN_SRC d static auto_heading_numbering_lv1 = ctRegex!(`^1~`, "m"); static auto_heading_numbering_lv2 = ctRegex!(`^2~`, "m"); @@ -276,7 +307,7 @@ static auto_heading_numbering_off_lv4 = ctRegex!(`^4~\S*?-\s`, "m ** no object_number object :ocn:off:object: -#+NAME: meta_rgx +#+NAME: meta_rgx_15 #+BEGIN_SRC d /+ no object_number object +/ static object_number_off = ctRegex!(`~#[ ]*$`, "m"); @@ -287,7 +318,7 @@ static repeated_character_line_separator = ctRegex!(`^(?:[ ]*(?:(?:[. ** no object_number block :ocn:off:block: -#+NAME: meta_rgx +#+NAME: meta_rgx_16 #+BEGIN_SRC d /+ no object_number block +/ static object_number_off_block = ctRegex!(`^--~#$`); @@ -298,7 +329,7 @@ static object_number_block_marks = ctRegex!(`^--[+~-]#$`); ** ignore outside code blocks :block:code: -#+NAME: meta_rgx +#+NAME: meta_rgx_17 #+BEGIN_SRC d /+ ignore outside code blocks +/ static skip_from_regular_parse = ctRegex!(`^(--[+~-]#|-[\\]{2}-|=[.\\]{2}=)$`); @@ -306,7 +337,7 @@ static skip_from_regular_parse = ctRegex!(`^(--[+~-]#|-[\\]{2}-|=[.\\]{2}=)$` ** line & page breaks :break: -#+NAME: meta_rgx +#+NAME: meta_rgx_18 #+BEGIN_SRC d /+ line & page breaks +/ static break_string = ctRegex!(`』`); @@ -314,7 +345,7 @@ static break_string = ctRegex!(`』`); ** biblio tags :biblio:tags: -#+NAME: meta_rgx +#+NAME: meta_rgx_19 #+BEGIN_SRC d /+ biblio tags +/ static biblio_tags = ctRegex!(`^(is|au|author_raw|author|author_arr|editor_raw|ed|editor_arr|ti|title|subtitle|fulltitle|lng|language|trans|src|jo|journal|in|vol|volume|edn|edition|yr|year|pl|place|pb|pub|publisher|url|pg|pages|note|short_name|id):\s+(.+)`); @@ -323,7 +354,7 @@ static biblio_abbreviations = ctRegex!(`^(au|ed|ti|lng ** bookindex split :bookindex:split: -#+NAME: meta_rgx +#+NAME: meta_rgx_20 #+BEGIN_SRC d /+ bookindex split +/ static bi_main_terms_split = ctRegex!(`\s*;\s*`); @@ -334,7 +365,7 @@ static bi_term_and_object_numbers_match = ctRegex!(`^(.+?)\+(\d+)` ** topic register split (document classify) -#+NAME: meta_rgx +#+NAME: meta_rgx_21 #+BEGIN_SRC d static topic_register_main_terms_split = ctRegex!(`\s*;\s*`); static topic_register_main_term_plus_rest_split = ctRegex!(`\s*:\s*`); @@ -344,7 +375,7 @@ static topic_register_multiple_sub_terms_split = ctRegex!(`␣([^|␣]+(? ** language codes :language:codes: -#+NAME: meta_rgx +#+NAME: meta_rgx_22 #+BEGIN_SRC d /+ language codes +/ auto language_code_and_filename = @@ -367,8 +398,17 @@ http://dlang.org/phobos/std_regex.html module doc_reform.io_out.rgx; static template spineRgxOut() { static struct RgxO { - <<prgmkup_rgx>> - <<sp_ch_xhtml_rgx>> + <<prgmkup_rgx_0>> + <<prgmkup_rgx_1>> + <<prgmkup_rgx_2>> + <<prgmkup_rgx_3>> + <<prgmkup_rgx_4>> + <<prgmkup_rgx_5>> + <<prgmkup_rgx_6>> + <<prgmkup_rgx_7>> + <<prgmkup_rgx_8>> + <<sp_ch_xhtml_rgx_0>> + <<sp_ch_xhtml_rgx_1>> } } #+END_SRC @@ -376,7 +416,7 @@ static template spineRgxOut() { ** special characters *** xhtml special characters -#+NAME: sp_ch_xhtml_rgx +#+NAME: sp_ch_xhtml_rgx_0 #+BEGIN_SRC d static xhtml_ampersand = ctRegex!(`[&]`, "m"); // & static xhtml_quotation = ctRegex!(`["]`, "m"); // " @@ -387,7 +427,7 @@ static xhtml_line_break = ctRegex!(` [\\]{2}`, "m"); / *** latex special characters -#+NAME: sp_ch_xhtml_rgx +#+NAME: sp_ch_xhtml_rgx_1 #+BEGIN_SRC d static latex_special_char = ctRegex!(`([%${}_#&\\])`); static latex_special_char_for_escape = ctRegex!(`([%${}_#\\])`); @@ -404,7 +444,7 @@ static latex_clean_bookindex_linebreak = ctRegex!(`\s*\\\\\\\\\s*`, " * 2. ctRegex defaults shared by meta & output (generic) ** misc generic -#+NAME: prgmkup_rgx +#+NAME: prgmkup_rgx_0 #+BEGIN_SRC d static newline = ctRegex!("\n", "mg"); static space = ctRegex!(`[ ]`, "mg"); @@ -415,7 +455,7 @@ static nbsp_char = ctRegex!(`░`, "mg"); ** filename (& path) (including insert file) :insert:file:path:filename: -#+NAME: prgmkup_rgx +#+NAME: prgmkup_rgx_1 #+BEGIN_SRC d static src_pth_sst_or_ssm = ctRegex!(`^(?P<path>[/]?(?:[a-zA-Z0-9._-]+/)*)(?P<filename>[a-zA-Z0-9._-]+[.](?P<extension>ss[tm]))$`); static src_pth_pod_sst_or_ssm = ctRegex!(`^(?P<podpath>[/]?(?:[a-zA-Z0-9._-]+/)*)media/text/[a-z]{2}/(?P<filename>[a-zA-Z0-9][a-zA-Z0-9._-]*?[.]ss[tm])$`); @@ -436,7 +476,7 @@ static src_formalised_file_path_parts = ctRegex!(`(?P<pth>(?:[/a *** inline breaks -#+NAME: prgmkup_rgx +#+NAME: prgmkup_rgx_2 #+BEGIN_SRC d /+ line breaks +/ static br_empty_line = ctRegex!(`\n[ ]*\n`, "mg"); @@ -447,7 +487,7 @@ static br_nl = ctRegex!(`┙`, "mg"); *** inline (internal program) markup footnotes endnotes :inline:footnote: -#+NAME: prgmkup_rgx +#+NAME: prgmkup_rgx_3 #+BEGIN_SRC d /+ inline markup footnotes endnotes +/ static inline_notes_al = ctRegex!(`【(?:[*+]\s+|\s*)(.+?)】`, "mg"); @@ -467,7 +507,7 @@ static inline_text_and_note_al_ = ctRegex!(`(.+?(?:【[*+] *** inline links -#+NAME: prgmkup_rgx +#+NAME: prgmkup_rgx_4 #+BEGIN_SRC d /+ inline markup links +/ static inline_image = ctRegex!(`(?P<pre>┥)☼(?P<imginf>(?P<img>[a-zA-Z0-9._-]+?\.(?:jpg|gif|png)),w(?P<width>\d+)h(?P<height>\d+))\s*(?P<post>.*?┝┤.*?├)`, "mg"); @@ -495,7 +535,7 @@ static quotation_mark_sql_insert_delimiter = ctRegex!("[']", "mg"); *** inline markup font face mod :inline:font:face: -#+NAME: prgmkup_rgx +#+NAME: prgmkup_rgx_5 #+BEGIN_SRC d /+ inline markup font face mod +/ static inline_mark_emphasis = ctRegex!(`(?P<mark>[*])\{(?P<text>.+?)\}[*]`, "mg"); @@ -510,7 +550,7 @@ static inline_mark_mono = ctRegex!(`(?P<mark>[#])\{( static inline_mark_cite = ctRegex!(`(?P<mark>["])\{(?P<text>.+?)\}["]`, "mg"); #+END_SRC -#+NAME: prgmkup_rgx +#+NAME: prgmkup_rgx_6 #+BEGIN_SRC d static inline_faces_line = ctRegex!(`^[*!/_]_ (?P<text>.+?)((?: [\\]{2}|[~]#){0,2}$)`); static inline_emphasis_line = ctRegex!(`^\*_ (?P<text>.+?)(?P<tail>(?: [\\]{2}|[~]#){0,2}$)`); @@ -519,7 +559,7 @@ static inline_italics_line = ctRegex!(`^/_ (?P<text>. static inline_underscore_line = ctRegex!(`^__ (?P<text>.+?)(?P<tail>(?: [\\]{2}|[~]#){0,2}$)`); #+END_SRC -#+NAME: prgmkup_rgx +#+NAME: prgmkup_rgx_7 #+BEGIN_SRC d /+ inline markup font face mod +/ static inline_emphasis = ctRegex!(`[*]┨(?P<text>.+?)┣[*]`, "mg"); @@ -538,7 +578,7 @@ static inline_fontface_clean = ctRegex!(`[*!_/^,+■‖ *** table related -#+NAME: prgmkup_rgx +#+NAME: prgmkup_rgx_8 #+BEGIN_SRC d /+ table delimiters +/ static table_delimiter_col = ctRegex!("[ ]*[┊][ ]*", "mg"); |