path: root/src/blocks.c

#include <stdlib.h>
#include <assert.h>
#include <stdio.h>
#include <stdbool.h>
#include <ctype.h>

#include "stmd.h"
#include "scanners.h"
#include "uthash.h"

static void finalize(block* b, int line_number);

static block* make_block(int tag, int start_line, int start_column)
{
    block* e;
    e = (block*) malloc(sizeof(block));
    e->tag = tag;
    e->open = true;
    e->last_line_blank = false;
    e->start_line = start_line;
    e->start_column = start_column;
    e->end_line = start_line;
    e->children = NULL;
    e->last_child = NULL;
    e->parent = NULL;
    e->top = NULL;
    e->attributes.refmap = NULL;
    gh_buf_init(&e->string_content, 32);
    e->string_pos = 0;
    e->inline_content = NULL;
    e->next = NULL;
    e->prev = NULL;
    return e;
}

// Create a root document block.
extern block* make_document()
{
    block * e = make_block(document, 1, 1);
    reference * map = NULL;
    reference ** refmap;
    refmap = (reference**) malloc(sizeof(reference*));
    *refmap = map;
    e->attributes.refmap = refmap;
    e->top = e;
    return e;
}

// Returns true if line has only space characters, else false.
bool is_blank(gh_buf *s, int offset)
{
    while (offset < s->size) {
        switch (s->ptr[offset]) {
            case '\n':
                return true;
            case ' ':
                offset++;
            default:
                return false;
        }
    }

    return true;
}

static inline bool can_contain(int parent_type, int child_type)
{
    return ( parent_type == document ||
            parent_type == block_quote ||
            parent_type == list_item ||
            (parent_type == list && child_type == list_item) );
}

static inline bool accepts_lines(int block_type)
{
    return (block_type == paragraph ||
            block_type == atx_header ||
            block_type == indented_code ||
            block_type == fenced_code);
}

static void add_line(block* block, gh_buf *ln, int offset)
{
    assert(block->open);
    gh_buf_put(&block->string_content, ln->ptr + offset, ln->size - offset);
}

static void remove_trailing_blank_lines(gh_buf *ln)
{
    int i;

    for (i = ln->size - 1; i >= 0; --i) {
        char c = ln->ptr[i];

        if (c != ' ' && c != '\t' && c != '\r' && c != '\n')
            break;
    }

    if (i < 0) {
        gh_buf_clear(ln);
        return;
    }

    i = gh_buf_strchr(ln, '\n', i);
    if (i >= 0)
        gh_buf_truncate(ln, i + 1);
}

// Check to see if a block ends with a blank line, descending
// if needed into lists and sublists.
static bool ends_with_blank_line(block* block)
{
    if (block->last_line_blank) {
        return true;
    }
    if ((block->tag == list || block->tag == list_item) && block->last_child) {
        return ends_with_blank_line(block->last_child);
    } else {
        return false;
    }
}

// Break out of all containing lists
static int break_out_of_lists(block ** bptr, int line_number)
{
    block * container = *bptr;
    block * b = container->top;
    // find first containing list:
    while (b && b->tag != list) {
        b = b->last_child;
    }
    if (b) {
        while (container && container != b) {
            finalize(container, line_number);
            container = container->parent;
        }
        finalize(b, line_number);
        *bptr = b->parent;
    }
    return 0;
}


static void finalize(block* b, int line_number)
{
    int firstlinelen;
    int pos;
    block* item;
    block* subitem;

    if (!b->open)
        return; // don't do anything if the block is already closed

    b->open = false;
    if (line_number > b->start_line) {
        b->end_line = line_number - 1;
    } else {
        b->end_line = line_number;
    }

    switch (b->tag) {
        case paragraph:
            pos = 0;
            while (gh_buf_at(&b->string_content, b->string_pos) == '[' &&
                    (pos = parse_reference(&b->string_content, b->string_pos,
                                           b->top->attributes.refmap))) {
                b->string_pos = pos;
            }
            if (is_blank(&b->string_content, b->string_pos)) {
                b->tag = reference_def;
            }
            break;

        case indented_code:
            remove_trailing_blank_lines(&b->string_content);
            gh_buf_putc(&b->string_content, '\n');
            break;

        case fenced_code:
            // first line of contents becomes info
            firstlinelen = gh_buf_strchr(&b->string_content, '\n', b->string_pos);
            gh_buf_set(
                &b->attributes.fenced_code_data.info,
                b->string_content.ptr + b->string_pos,
                firstlinelen
            );

            b->string_pos = firstlinelen + 1;

            gh_buf_trim(&b->attributes.fenced_code_data.info);
            unescape_buffer(&b->attributes.fenced_code_data.info);
            break;

        case list: // determine tight/loose status
            b->attributes.list_data.tight = true; // tight by default
            item = b->children;

            while (item) {
                // check for non-final non-empty list item ending with blank line:
                if (item->last_line_blank && item->next) {
                    b->attributes.list_data.tight = false;
                    break;
                }
                // recurse into children of list item, to see if there are
                // spaces between them:
                subitem = item->children;
                while (subitem) {
                    if (ends_with_blank_line(subitem) &&
                            (item->next || subitem->next)) {
                        b->attributes.list_data.tight = false;
                        break;
                    }
                    subitem = subitem->next;
                }
                if (!(b->attributes.list_data.tight)) {
                    break;
                }
                item = item->next;
            }

            break;

        default:
            break;
    }
}

// Add a block as child of another.  Return pointer to child.
extern block* add_child(block* parent,
        int block_type, int start_line, int start_column)
{
    assert(parent);

    // if 'parent' isn't the kind of block that can accept this child,
    // then back up til we hit a block that can.
    while (!can_contain(parent->tag, block_type)) {
        finalize(parent, start_line);
        parent = parent->parent;
    }

    block* child = make_block(block_type, start_line, start_column);
    child->parent = parent;
    child->top = parent->top;

    if (parent->last_child) {
        parent->last_child->next = child;
        child->prev = parent->last_child;
    } else {
        parent->children = child;
        child->prev = NULL;
    }
    parent->last_child = child;
    return child;
}

// Free a block list and any children.
extern void free_blocks(block* e)
{
    block * next;
    while (e != NULL) {
        next = e->next;
        free_inlines(e->inline_content);
        gh_buf_free(&e->string_content);
        if (e->tag == fenced_code) {
            gh_buf_free(&e->attributes.fenced_code_data.info);
        } else if (e->tag == document) {
            free_reference_map(e->attributes.refmap);
        }
        free_blocks(e->children);
        free(e);
        e = next;
    }
}

// Walk through block and all children, recursively, parsing
// string content into inline content where appropriate.
void process_inlines(block* cur, reference** refmap)
{
    switch (cur->tag) {
        case paragraph:
        case atx_header:
        case setext_header:
            cur->inline_content = parse_inlines(&cur->string_content, cur->string_pos, refmap);
            // MEM
            // gh_buf_free(&cur->string_content);
            break;

        default:
            break;
    }

    block *child = cur->children;
    while (child != NULL) {
        process_inlines(child, refmap);
        child = child->next;
    }
}

// Attempts to parse a list item marker (bullet or enumerated).
// On success, returns length of the marker, and populates
// data with the details.  On failure, returns 0.
static int parse_list_marker(gh_buf *ln, int pos,
        struct ListData ** dataptr)
{
    char c;
    int startpos;
    struct ListData * data;

    startpos = pos;
    c = gh_buf_at(ln, pos);

    if ((c == '*' || c == '-' || c == '+') && !scan_hrule(ln, pos)) {
        pos++;
        if (!isspace(gh_buf_at(ln, pos))) {
            return 0;
        }
        data = malloc(sizeof(struct ListData));
        data->marker_offset = 0; // will be adjusted later
        data->list_type = bullet;
        data->bullet_char = c;
        data->start = 1;
        data->delimiter = period;
        data->tight = false;

    } else if (isdigit(c)) {
        int start = 0;

        do {
            start = (10 * start) + (gh_buf_at(ln, pos) - '0');
            pos++;
        } while (isdigit(gh_buf_at(ln, pos)));

        c = gh_buf_at(ln, pos);
        if (c == '.' || c == ')') {
            pos++;
            if (!isspace(gh_buf_at(ln, pos))) {
                return 0;
            }
            data = malloc(sizeof(struct ListData));
            data->marker_offset = 0; // will be adjusted later
            data->list_type = ordered;
            data->bullet_char = 0;
            data->start = start;
            data->delimiter = (c == '.' ? period : parens);
            data->tight = false;
        } else {
            return 0;
        }

    } else {
        return 0;
    }

    *dataptr = data;
    return (pos - startpos);
}

// Return 1 if list item belongs in list, else 0.
static int lists_match(struct ListData list_data,
        struct ListData item_data)
{
    return (list_data.list_type == item_data.list_type &&
            list_data.delimiter == item_data.delimiter &&
            // list_data.marker_offset == item_data.marker_offset &&
            list_data.bullet_char == item_data.bullet_char);
}

static void expand_tabs(gh_buf *ob, const unsigned char *line, size_t size)
{
    size_t  i = 0, tab = 0;

    while (i < size) {
        size_t org = i;

        while (i < size && line[i] != '\t') {
            i++; tab++;
        }

        if (i > org)
            gh_buf_put(ob, line + org, i - org);

        if (i >= size)
            break;

        do {
            gh_buf_putc(ob, ' '); tab++;
        } while (tab % 4);

        i++;
    }
}

static block *finalize_parsing(block *document, int linenum)
{
    while (document != document->top) {
        finalize(document, linenum);
        document = document->parent;
    }

    finalize(document, linenum);
    process_inlines(document, document->attributes.refmap);

    return document;
}

extern block *stmd_parse_file(FILE *f)
{
    gh_buf line = GH_BUF_INIT;
    unsigned char buffer[4096];
    int linenum = 1;
    block *document = make_document();

    while (fgets((char *)buffer, sizeof(buffer), f)) {
        expand_tabs(&line, buffer, strlen(buffer));
        incorporate_line(&line, linenum, &document);
        gh_buf_clear(&line);
        linenum++;
    }

    gh_buf_free(&line);
    return finalize_document(document, linenum);
}

extern block *stmd_parse_document(const unsigned char *buffer, size_t len)
{
    gh_buf line = GH_BUF_INIT;
    int linenum = 1;
    const unsigned char *end = buffer + len;
    block *document = make_document();

    while (buffer < end) {
        const char *eol = memchr(buffer, '\n', end - buffer);

        if (!eol) {
            expand_tabs(&line, buffer, end - buffer);
            buffer = end;
        } else {
            expand_tabs(&line, buffer, (eol - buffer) + 1);
            buffer += (eol - buffer) + 1;
        }

        incorporate_line(&line, linenum, &document);
        gh_buf_clear(&line);
        linenum++;
    }

    gh_buf_free(&line);
    return finalize_document(document, linenum);
}

// Process one line at a time, modifying a block.
// Returns 0 if successful.  curptr is changed to point to
// the currently open block.
extern void incorporate_line(gh_buf *ln, int line_number, block** curptr)
{
    block* last_matched_container;
    int offset = 0;
    int matched = 0;
    int lev = 0;
    int i;
    struct ListData * data = NULL;
    bool all_matched = true;
    block* container;
    block* cur = *curptr;
    bool blank = false;
    int first_nonspace;
    int indent;

    // container starts at the document root.
    container = cur->top;

    // for each containing block, try to parse the associated line start.
    // bail out on failure:  container will point to the last matching block.

    while (container->last_child && container->last_child->open) {
        container = container->last_child;

        first_nonspace = offset;
        while (gh_buf_at(ln, first_nonspace) == ' ') {
            first_nonspace++;
        }

        indent = first_nonspace - offset;
        blank = gh_buf_at(ln, first_nonspace) == '\n';

        if (container->tag == block_quote) {

            matched = indent <= 3 && gh_buf_at(ln, first_nonspace) == '>';
            if (matched) {
                offset = first_nonspace + 1;
                if (gh_buf_at(ln, offset) == ' ') {
                    offset++;
                }
            } else {
                all_matched = false;
            }

        } else if (container->tag == list_item) {

            if (indent >= container->attributes.list_data.marker_offset +
                    container->attributes.list_data.padding) {
                offset += container->attributes.list_data.marker_offset +
                    container->attributes.list_data.padding;
            } else if (blank) {
                offset = first_nonspace;
            } else {
                all_matched = false;
            }

        } else if (container->tag == indented_code) {

            if (indent >= CODE_INDENT) {
                offset += CODE_INDENT;
            } else if (blank) {
                offset = first_nonspace;
            } else {
                all_matched = false;
            }

        } else if (container->tag == atx_header ||
                container->tag == setext_header) {

            // a header can never contain more than one line
            all_matched = false;

        } else if (container->tag == fenced_code) {

            // skip optional spaces of fence offset
            i = container->attributes.fenced_code_data.fence_offset;
            while (i > 0 && gh_buf_at(ln, offset) == ' ') {
                offset++;
                i--;
            }

        } else if (container->tag == html_block) {

            if (blank) {
                all_matched = false;
            }

        } else if (container->tag == paragraph) {

            if (blank) {
                container->last_line_blank = true;
                all_matched = false;
            }

        }

        if (!all_matched) {
            container = container->parent;  // back up to last matching block
            break;
        }
    }

    last_matched_container = container;

    // check to see if we've hit 2nd blank line, break out of list:
    if (blank && container->last_line_blank) {
        break_out_of_lists(&container, line_number);
    }

    // unless last matched container is code block, try new container starts:
    while (container->tag != fenced_code && container->tag != indented_code &&
            container->tag != html_block) {

        first_nonspace = offset;
        while (gh_buf_at(ln, first_nonspace) == ' ') {
            first_nonspace++;
        }

        indent = first_nonspace - offset;
        blank = gh_buf_at(ln, first_nonspace) == '\n';

        if (indent >= CODE_INDENT) {

            if (cur->tag != paragraph && !blank) {
                offset += CODE_INDENT;
                container = add_child(container, indented_code, line_number, offset + 1);
            } else { // indent > 4 in lazy line
                break;
            }

        } else if (gh_buf_at(ln, first_nonspace) == '>') {

            offset = first_nonspace + 1;
            // optional following character
            if (gh_buf_at(ln, offset) == ' ') {
                offset++;
            }
            container = add_child(container, block_quote, line_number, offset + 1);

        } else if ((matched = scan_atx_header_start(ln, first_nonspace))) {

            offset = first_nonspace + matched;
            container = add_child(container, atx_header, line_number, offset + 1);

            int hashpos = gh_buf_strchr(ln, '#', first_nonspace);
            assert(hashpos >= 0);

            int level = 0;
            while (gh_buf_at(ln, hashpos) == '#') {
                level++;
                hashpos++;
            }
            container->attributes.header_level = level;

        } else if ((matched = scan_open_code_fence(ln, first_nonspace))) {

            container = add_child(container, fenced_code, line_number,
                    first_nonspace + 1);
            container->attributes.fenced_code_data.fence_char = gh_buf_at(ln,
                    first_nonspace);
            container->attributes.fenced_code_data.fence_length = matched;
            container->attributes.fenced_code_data.fence_offset =
                first_nonspace - offset;
            offset = first_nonspace + matched;

        } else if ((matched = scan_html_block_tag(ln, first_nonspace))) {

            container = add_child(container, html_block, line_number,
                    first_nonspace + 1);
            // note, we don't adjust offset because the tag is part of the text

        } else if (container->tag == paragraph &&
                (lev = scan_setext_header_line(ln, first_nonspace)) &&
                // check that there is only one line in the paragraph:
                gh_buf_strrchr(&container->string_content, '\n',
                    gh_buf_len(&container->string_content) - 2) < 0) {

            container->tag = setext_header;
            container->attributes.header_level = lev;
            offset = gh_buf_len(ln) - 1;

        } else if (!(container->tag == paragraph && !all_matched) &&
                (matched = scan_hrule(ln, first_nonspace))) {

            // it's only now that we know the line is not part of a setext header:
            container = add_child(container, hrule, line_number, first_nonspace + 1);
            finalize(container, line_number);
            container = container->parent;
            offset = gh_buf_len(ln) - 1;

        } else if ((matched = parse_list_marker(ln, first_nonspace, &data))) {

            // compute padding:
            offset = first_nonspace + matched;
            i = 0;
            while (i <= 5 && gh_buf_at(ln, offset + i) == ' ') {
                i++;
            }
            // i = number of spaces after marker, up to 5
            if (i >= 5 || i < 1 || gh_buf_at(ln, offset) == '\n') {
                data->padding = matched + 1;
                if (i > 0) {
                    offset += 1;
                }
            } else {
                data->padding = matched + i;
                offset += i;
            }

            // check container; if it's a list, see if this list item
            // can continue the list; otherwise, create a list container.

            data->marker_offset = indent;

            if (container->tag != list ||
                    !lists_match(container->attributes.list_data, *data)) {
                container = add_child(container, list, line_number,
                        first_nonspace + 1);
                container->attributes.list_data = *data;
            }

            // add the list item
            container = add_child(container, list_item, line_number,
                    first_nonspace + 1);
            container->attributes.list_data = *data;
            free(data);

        } else {
            break;
        }

        if (accepts_lines(container->tag)) {
            // if it's a line container, it can't contain other containers
            break;
        }
    }

    // what remains at offset is a text line.  add the text to the
    // appropriate container.

    first_nonspace = offset;
    while (gh_buf_at(ln, first_nonspace) == ' ') {
        first_nonspace++;
    }

    indent = first_nonspace - offset;
    blank = gh_buf_at(ln, first_nonspace) == '\n';

    // block quote lines are never blank as they start with >
    // and we don't count blanks in fenced code for purposes of tight/loose
    // lists or breaking out of lists.  we also don't set last_line_blank
    // on an empty list item.
    container->last_line_blank = (blank &&
            container->tag != block_quote &&
            container->tag != fenced_code &&
            !(container->tag == list_item &&
                container->children == NULL &&
                container->start_line == line_number));

    block *cont = container;
    while (cont->parent) {
        cont->parent->last_line_blank = false;
        cont = cont->parent;
    }

    if (cur != last_matched_container &&
            container == last_matched_container &&
            !blank &&
            cur->tag == paragraph &&
            gh_buf_len(&cur->string_content) > 0) {

        add_line(cur, ln, offset);

    } else { // not a lazy continuation

        // finalize any blocks that were not matched and set cur to container:
        while (cur != last_matched_container) {

            finalize(cur, line_number);
            cur = cur->parent;
            assert(cur != NULL);
        }

        if (container->tag == indented_code) {

            add_line(container, ln, offset);

        } else if (container->tag == fenced_code) {

            matched = (indent <= 3
                    && gh_buf_at(ln, first_nonspace) == container->attributes.fenced_code_data.fence_char)
                && scan_close_code_fence(ln, first_nonspace,
                        container->attributes.fenced_code_data.fence_length);
            if (matched) {
                // if closing fence, don't add line to container; instead, close it:
                finalize(container, line_number);
                container = container->parent; // back up to parent
            } else {
                add_line(container, ln, offset);
            }

        } else if (container->tag == html_block) {

            add_line(container, ln, offset);

        } else if (blank) {

            // ??? do nothing

        } else if (container->tag == atx_header) {
            // chop off trailing ###s...use a scanner?
            gh_buf_trim(ln);
            int p = gh_buf_len(ln) - 1;

            // if string ends in #s, remove these:
            while (gh_buf_at(ln, p) == '#') {
                p--;
            }
            if (gh_buf_at(ln, p) == '\\') {
                // the last # was escaped, so we include it.
                p++;
            }

            gh_buf_truncate(ln, p + 1);
            add_line(container, ln, first_nonspace);
            finalize(container, line_number);
            container = container->parent;

        } else if (accepts_lines(container->tag)) {

            add_line(container, ln, first_nonspace);

        } else if (container->tag != hrule && container->tag != setext_header) {

            // create paragraph container for line
            container = add_child(container, paragraph, line_number, first_nonspace + 1);
            add_line(container, ln, first_nonspace);

        } else {
            assert(false);
        }

        *curptr = container;
    }
}