1 # mdcode: extract C code from a _markdown_ file.
3 _markdown_ is a popular format for simple text markup which can easily
4 be converted to HTML. As it allows easy indication of sections of
5 code, it is quite suitable for use in literate programming. This file
6 is an example of that usage.
8 The code included below provides two related functionalities.
9 Firstly it provides a library routine for extracting code out of a
10 _markdown_ file, so that other routines might make use of it.
12 Secondly it provides a simple client of this routine which extracts
13 1 or more C-language files from a markdown document so they can be
14 passed to a C compiler. These two combined to make a tool that is needed
15 to compile this tool. Yes, this is circular. A prototype tool was
16 used for the first extraction.
18 The tool provided is described as specific to the C language as it
21 ##### Example: a _line_ command
23 #line __line-number__ __file-name__
25 lines so that the C compiler will report where in the markdown file
26 any error is found. This tool is suitable for any other language
27 which allows the same directive, or will treat it as a comment.
31 Literate programming is more than just including comments with the
32 code, even nicely formatted comments. It also involves presenting the
33 code in an order that makes sense to a human, rather than an order
34 that makes sense to a compiler. For this reason a core part of any
35 literate programming tool is the ability to re-arrange the code found
36 in the document into a different order in the final code file - or
37 files. This requires some form of linkage to be encoded.
39 The approach taken here is focused around section headings - of any
42 All the code in any section is treated as a single sequential
43 collection of code, and is named by the section that it is in. If
44 multiple sections have the same name, then the code blocks in all of
45 them are joined together in the order they appear in the document.
47 A code section can contain a special marker which starts with 2
49 The text after the marker must be the name of some section which
50 contains code. Code from that section will be interpolated in place
51 of the marker, and will be indented to match the indent of the marker.
53 It is not permitted for the same code to be interpolated multiple
54 times. Allowing this might make some sense, but it is probably a
55 mistake, and prohibiting it make some of the code a bit cleaner.
57 Equally, every section of code should be interpolated at least once -
58 with one exception. This exception is imposed by the
59 tool, not the library. A different client could impose different
60 rules on the names of top-level code sections.
62 One example of the exception we have already seen. A section name
63 starting __Example:__ indicates code that is not to be included in the
64 final product. Any leading word will do, providing there is a space,
65 and the first space is preceded by a colon, that section name will be
68 A special case of this exception exists for the leading word
69 __File__. These sections are the top level code sections and they
70 will be written to the named file. Thus a section named
71 __File: foo__ should not be referenced by another section, and its
72 contents after all references are expanded will be written to the file
75 Any section containing code that does not start __Word:__
76 must be included in some other section exactly once.
80 Allowing multiple top level code sections which name different files
81 means that one _markdown_ document can describe several files. This
82 is very useful with the C language where a program file and a header
83 file might be related. For the present document we will have a header
84 file and two code files, one with the library content and one for the
87 It will also be very convenient to create a `makefile` fragment to
88 ensure the code is compiled correctly. A simple `make -f mdcode.mk`
89 will "do the right thing".
95 mdcode.h libmdcode.c md2c.c mdcode.mk : mdcode.mdc
103 ## exported functions
105 ### File: libmdcode.c
114 ## internal functions
119 libmdcode.o : libmdcode.c mdcode.h
120 $(CC) $(CFLAGS) -c libmdcode.c
136 md2c : md2c.o libmdcode.o
137 $(CC) $(CFLAGS) -o md2c md2c.o libmdcode.o
138 md2c.o : md2c.c mdcode.h
139 $(CC) $(CFLAGS) -c md2c.c
143 As the core purpose of _mdcode_ is to discover and re-arrange blocks
144 of text, it makes sense to map the whole document file into memory and
145 produce a data structure which lists various parts of the file in the
146 appropriate order. Each node in this structure will have some text
147 from the document, a child pointer, and a next pointer, any of which
148 might not be present. The text is most easily stored as a pointer and a
149 length. We'll call this a `text`
151 A list of these `code_nodes` will belong to each section and it will
152 be useful to have a separate `section` data structure to store the
153 list of `code_nodes`, the section name, and some other information.
155 This other information will include a reference counter so we can
156 ensure proper referencing, and an `indent` depth. As referenced
157 content can have an extra indent added, we need to know what that is.
158 The `code_node` will also have an `indent` depth which eventually gets
159 set to the sum for the indents from all references on the path from
162 Finally we need to know if the `code_node` was recognised by being
163 indented or not. If it was, the client of this data will want to
164 strip of the leading tab or 4 spaces. Hence a `needs_strip` flag is
176 struct code_node *code;
177 struct section *next;
185 struct code_node *next;
186 struct section *child;
193 struct code_node *last;
198 You will note that the `struct psection` contains an anonymous `struct
199 section` embedded at the start. To make this work right, GCC
200 requires the `-fplan9-extensions` flag.
202 ##### File: mdcode.mk
204 CFLAGS += -fplan9-extensions
206 ### Manipulating the node
208 Though a tree with `next` and `child` links is the easiest way to
209 assemble the various code sections, it is not the easiest form for
210 using them. For that a simple list would be best.
212 So once we have a fully linked File section we will want to linearize
213 it, so that the `child` links become `NULL` and the `next` links will
214 find everything required. It is at this stage that the requirements
215 that each section is linked only once becomes import.
217 `code_linearize` will merge the `code_node`s from any child into the
218 given `code_node`. As it does this it sets the 'indent' field for
221 Note that we don't clear the section's `last` pointer, even though
222 it no longer owns any code. This allows subsequent code to see if a
223 section ever had any code, and to report an error if a section is
224 referenced but not defined.
226 ##### internal functions
228 static void code_linearize(struct code_node *code)
231 for (t = code; t; t = t->next)
233 for (; code; code = code->next)
235 struct code_node *next = code->next;
236 struct psection *pchild =
237 (struct psection *)code->child;
238 int indent = pchild->indent;
239 code->next = code->child->code;
240 code->child->code = NULL;
242 for (t = code; t->next; t = t->next)
243 t->next->indent = code->indent + indent;
248 Once a client has made use of a linearized code set, it will probably
251 void code_free(struct code_node *code)
254 struct code_node *this;
256 code_linearize(code);
263 ##### exported functions
265 void code_free(struct code_node *code);
267 ### Building the tree
269 As we parse the document there are two things we will want to do to
270 node trees: add some text or add a reference. We'll assume for now
271 that the relevant section structures have been found, and will just
272 deal with the `code_node`.
274 Adding text simply means adding another node. We will never have
275 empty nodes, even if the last node only has a child, new text must go
278 ##### internal functions
280 static void code_add_text(struct psection *where, struct text txt,
281 int line_no, int needs_strip)
286 n = malloc(sizeof(*n));
289 n->line_no = line_no;
290 n->needs_strip = needs_strip;
294 where->last->next = n;
300 However when adding a link, we might be able to include it in the last
301 `code_node` if it currently only has text.
303 void code_add_link(struct psection *where, struct psection *to,
309 to->refcnt++; // this will be checked elsewhere
310 if (where->last && where->last->child == NULL) {
311 where->last->child = to;
314 n = malloc(sizeof(*n));
321 where->last->next = n;
329 Now we need a lookup table to be able to find sections by name.
330 Something that provides an `n*log(N)` search time is probably
331 justified, but for now I want a minimal stand-alone program so a
332 linked list managed by insertion-sort will do.
334 The text compare function will likely be useful for any clients of our
335 library, so we may as well export it.
337 If we cannot find a section, we simply want to create it. This allows
338 sections and references to be created in any order. Sections with
339 no references or no content will cause a warning eventually.
341 #### exported functions
343 int text_cmp(struct text a, struct text b);
345 #### internal functions
347 int text_cmp(struct text a, struct text b)
352 int cmp = strncmp(a.txt, b.txt, len);
356 return a.len - b.len;
359 static struct psection *section_find(struct psection **list, struct text name)
361 struct psection *new;
363 int cmp = text_cmp((*list)->section, name);
368 list = (struct psection **)&((*list)->next);
370 /* Add this section */
371 new = malloc(sizeof(*new));
382 ## Parsing the _markdown_
384 Parsing markdown is fairly easy, though there are complications.
386 The document is divided into "paragraphs" which are mostly separated by blank
387 lines (which may contain white space). The first few characters of
388 the first line of a paragraph determine the type of paragraph. For
389 our purposes we are only interested in list paragraphs, code
390 paragraphs, section headings, and everything else. Section headings
391 are single-line paragraphs and so do not require a preceding or
392 following blank line.
394 Section headings start with 1 or more hash characters (__#__). List
395 paragraphs start with hyphen, asterisk, plus, or digits followed by a
396 period. Code paragraphs aren't quite so easy.
398 The "standard" code paragraph starts with 4 or more spaces, or a tab.
399 However if the previous paragraph was a list paragraph, then those
400 spaces indicate another paragraph in the same list item, and 8 or
401 more spaces are required. Unless a nested list is in effect, in
402 which case 12 or more are need. Unfortunately not all _markdown_
403 parsers agree on nested lists.
405 Two alternate styles for marking code are in active use. "Github" uses
406 three backticks(_`` ``` ``_), while "pandoc" uses three or more tildes
407 (_~~~_). In these cases the code should not be indented.
409 Trying to please everyone as much as possible, this parser will handle
410 everything except for code inside lists.
412 So an indented (4+) paragraph after a list paragraph is always a list
413 paragraph, otherwise it is a code paragraph. A paragraph that starts
414 with three backticks or three tildes is code which continues until a
415 matching string of backticks or tildes.
419 While walking the document looking for various markers we will *not*
420 use the `struct text` introduced earlier as advancing that requires
421 updating both start and length which feels clumsy. Instead we will
422 carry `pos` and `end` pointers, only the first of which needs to
425 So to start, we need to skip various parts of the document. `lws`
426 stands for "Linear White Space" and is a term that comes from the
427 Email RFCs (e.g. RFC822). `line` and `para` are self explanatory.
428 Note that `skip_para` needs to update the current line number.
429 `skip_line` doesn't but every caller should.
431 #### internal functions
433 static char *skip_lws(char *pos, char *end)
435 while (pos < end && (*pos == ' ' || *pos == '\t'))
440 static char *skip_line(char *pos, char *end)
442 while (pos < end && *pos != '\n')
449 static char *skip_para(char *pos, char *end, int *line_no)
451 /* Might return a pointer to a blank line, as only
452 * one trailing blank line is skipped
455 pos = skip_line(pos, end);
461 *(pos = skip_lws(pos, end)) != '\n') {
462 pos = skip_line(pos, end);
465 if (pos < end && *pos == '\n') {
472 ### Recognising things
474 Recognising a section header is trivial and doesn't require a
475 function. However we need to extract the content of a section header
476 as a `struct text` for passing to `section_find`.
477 Recognising the start of a new list is fairly easy. Recognising the
478 start (and end) of code is a little messy so we provide a function for
479 matching the first few characters, which has a special case for "4
482 #### internal includes
487 #### internal functions
489 static struct text take_header(char *pos, char *end)
493 while (pos < end && *pos == '#')
495 while (pos < end && *pos == ' ')
498 while (pos < end && *pos != '\n')
500 while (pos > section.txt &&
501 (pos[-1] == '#' || pos[-1] == ' '))
503 section.len = pos - section.txt;
507 static int is_list(char *pos, char *end)
509 if (strchr("-*+", *pos))
512 while (pos < end && isdigit(*pos))
514 if (pos < end && *pos == '.')
520 static int matches(char *start, char *pos, char *end)
523 return matches("\t", pos, end) ||
524 matches(" ", pos, end);
525 return (pos + strlen(start) < end &&
526 strncmp(pos, start, strlen(start)) == 0);
529 ### Extracting the code
531 Now that we can skip paragraphs and recognise what type each paragraph
532 is, it is time to parse the file and extract the code. We'll do this
533 in two parts, first we look at what to do with some code once we
534 find it, and then how to actually find it.
536 When we have some code, we know where it is, what the end marker
537 should look like, and which section it is in.
539 There are two sorts of end markers: the presence of a particular
540 string, or the absence of an indent. We will use a string to
541 represent a presence, and a `NULL` to represent the absence.
543 While looking at code we don't think about paragraphs at all - just
544 look for a line that starts with the right thing.
545 Every line that is still code then needs to be examined to see if it
546 is a section reference.
548 When a section reference is found, all preceding code (if any) must be
549 added to the current section, then the reference is added.
551 When we do find the end of the code, all text that we have found but
552 not processed needs to be saved too.
554 When adding a reference we need to set the `indent`. This is the
555 number of spaces (counting 8 for tabs) after the natural indent of the
556 code (which is a tab or 4 spaces). We use a separate function `count_spaces`
559 If there are completely blank linkes (no indent) at the end of the found code,
560 these should be considered to be spacing between the code and the next section,
561 and so no included in the code. When a marker is used to explicitly mark the
562 end of the code, we don't need to check for these blank lines.
564 #### internal functions
566 static int count_space(char *sol, char *p)
579 static char *take_code(char *pos, char *end, char *marker,
580 struct psection **table, struct text section,
584 int line_no = *line_nop;
585 int start_line = line_no;
586 struct psection *sect;
588 sect = section_find(table, section);
594 if (marker && matches(marker, pos, end))
597 (skip_lws(pos, end))[0] != '\n' &&
598 !matches(NULL, pos, end))
599 /* Paragraph not indented */
602 /* Still in code - check for reference */
607 else if (strcmp(sol, " ") == 0)
610 t = skip_lws(sol, end);
611 if (t[0] != '#' || t[1] != '#') {
612 /* Just regular code here */
613 pos = skip_line(sol, end);
621 txt.len = pos - start;
622 code_add_text(sect, txt, start_line,
625 ref = take_header(t, end);
627 struct psection *refsec = section_find(table, ref);
628 code_add_link(sect, refsec, count_space(sol, t));
630 pos = skip_line(t, end);
633 start_line = line_no;
638 txt.len = pos - start;
639 /* strip trailing blank lines */
640 while (!marker && txt.len > 2 &&
641 start[txt.len-1] == '\n' &&
642 start[txt.len-2] == '\n')
645 code_add_text(sect, txt, start_line,
649 pos = skip_line(pos, end);
658 It is when looking for the code that we actually use the paragraph
659 structure. We need to recognise section headings so we can record the
660 name, list paragraphs so we can ignore indented follow-on paragraphs,
661 and the three different markings for code.
663 #### internal functions
665 static struct psection *code_find(char *pos, char *end)
667 struct psection *table = NULL;
670 struct text section = {0};
674 section = take_header(pos, end);
676 pos = skip_line(pos, end);
678 } else if (is_list(pos, end)) {
680 pos = skip_para(pos, end, &line_no);
681 } else if (!in_list && matches(NULL, pos, end)) {
682 pos = take_code(pos, end, NULL, &table,
684 } else if (matches("```", pos, end)) {
686 pos = skip_line(pos, end);
688 pos = take_code(pos, end, "```", &table,
690 } else if (matches("~~~", pos, end)) {
692 pos = skip_line(pos, end);
694 pos = take_code(pos, end, "~~~", &table,
699 pos = skip_para(pos, end, &line_no);
705 ### Returning the code
707 Having found all the code blocks and gathered them into a list of
708 section, we are now ready to return them to the caller. This is where
709 to perform consistency checks, like at most one reference and at least
710 one definition for each section.
712 All the sections with no references are returned in a list for the
713 caller to consider. The are linearized first so that the substructure
714 is completely hidden -- except for the small amount of structure
715 displayed in the line numbers.
717 To return errors, we have the caller pass a function which takes an
718 error message - a `code_err_fn`.
722 typedef void (*code_err_fn)(char *msg);
724 #### internal functions
725 struct section *code_extract(char *pos, char *end, code_err_fn error)
727 struct psection *table;
728 struct section *result = NULL;
729 struct section *tofree = NULL;
731 table = code_find(pos, end);
734 struct psection *t = (struct psection*)table->next;
735 if (table->last == NULL) {
738 "Section \"%.*s\" is referenced but not declared",
739 table->section.len, table->section.txt);
743 if (table->refcnt == 0) {
744 /* Root-section, return it */
745 table->next = result;
747 code_linearize(result->code);
749 table->next = tofree;
751 if (table->refcnt > 1) {
754 "Section \"%.*s\" referenced multiple times (%d).",
755 table->section.len, table->section.txt,
764 struct section *t = tofree->next;
771 ##### exported functions
773 struct section *code_extract(char *pos, char *end, code_err_fn error);
777 Now that we can extract code from a document and link it all together
778 it is time to do something with that code. Firstly we need to print
781 ### Printing the Code
783 Printing is mostly straight forward - we just walk the list and print
784 the code sections, adding whatever indent is required for each line.
785 However there is a complication (isn't there always)?
787 For code that was recognised because the paragraph was indented, we
788 need to strip that indent first. For other code, we don't.
790 The approach taken here is simple, though it could arguably be wrong
791 in some unlikely cases. So it might need to be fixed later.
793 If the first line of a code block is indented, then either one tab or
794 4 spaces are striped from every non-blank line.
796 This could go wrong if the first line of a code block marked by
797 _`` ``` ``_ is indented. To overcome this we would need to
798 record some extra state in each `code_node`. For now we won't bother.
800 The indents we insert will mostly be spaces. All-spaces doesn't work
801 for `Makefiles`, so if the indent is 8 or more, we use a TAB first.
803 ##### internal functions
805 void code_node_print(FILE *out, struct code_node *node,
808 for (; node; node = node->next) {
809 char *c = node->code.txt;
810 int len = node->code.len;
815 fprintf(out, "#line %d \"%s\"\n",
816 node->line_no, fname);
818 if (node->indent >= 8)
819 fprintf(out, "\t%*s", node->indent - 8, "");
821 fprintf(out, "%*s", node->indent, "");
822 if (node->needs_strip) {
823 if (*c == '\t' && len > 1) {
826 } else if (strncmp(c, " ", 4) == 0 && len > 4) {
835 } while (len && c[-1] != '\n');
840 ###### exported functions
841 void code_node_print(FILE *out, struct code_node *node, char *fname);
843 ### Bringing it all together
845 We are just about ready for the `main` function of the tool which will
846 extract all this lovely code and compile it. Just one helper is still
849 #### Handling filenames
851 Section names are stored in `struct text` which is not `nul`
852 terminated. Filenames passed to `open` need to be null terminated.
853 So we need to convert one to the other, and strip the leading `File:`
854 of while we are at it.
856 ##### client functions
858 static void copy_fname(char *name, int space, struct text t)
863 if (len < 5 || strncmp(sec, "File:", 5) != 0)
867 while (len && sec[0] == ' ') {
873 strncpy(name, sec, len);
879 And now we take a single file name, extract the code, and if there are
880 no error we write out a file for each appropriate code section. And
883 ##### client includes
887 #include <sys/mman.h>
890 ##### client functions
893 static void pr_err(char *msg)
896 fprintf(stderr, "%s\n", msg);
899 static char *strnchr(char *haystack, int len, char needle)
901 while (len > 0 && *haystack && *haystack != needle) {
905 return len > 0 && *haystack == needle ? haystack : NULL;
908 int main(int argc, char *argv[])
913 struct text section = {NULL, 0};
914 struct section *table, *s, *prev;
917 if (argc != 2 && argc != 3) {
918 fprintf(stderr, "Usage: mdcode file.mdc [section]\n");
922 section.txt = argv[2];
923 section.len = strlen(argv[2]);
926 fd = open(argv[1], O_RDONLY);
928 fprintf(stderr, "mdcode: cannot open %s: %s\n",
929 argv[1], strerror(errno));
932 len = lseek(fd, 0, 2);
933 file = mmap(NULL, len, PROT_READ, MAP_SHARED, fd, 0);
934 table = code_extract(file, file+len, pr_err);
937 (code_free(s->code), prev = s, s = s->next, free(prev))) {
940 char *spc = strnchr(s->section.txt, s->section.len, ' ');
942 if (spc > s->section.txt && spc[-1] == ':') {
943 if (strncmp(s->section.txt, "File: ", 6) != 0 &&
944 (section.txt == NULL ||
945 text_cmp(s->section, section) != 0))
946 /* Ignore this section */
949 fprintf(stderr, "Code in unreferenced section that is not ignored or a file name: %.*s\n",
950 s->section.len, s->section.txt);
955 if (text_cmp(s->section, section) == 0)
956 code_node_print(stdout, s->code, argv[1]);
959 copy_fname(fname, sizeof(fname), s->section);
961 fprintf(stderr, "Missing file name at:%.*s\n",
962 s->section.len, s->section.txt);
966 fl = fopen(fname, "w");
968 fprintf(stderr, "Cannot create %s: %s\n",
969 fname, strerror(errno));
973 code_node_print(fl, s->code, argv[1]);