From 5c347a95a0e25f9a52688ec81c7052bd131e1d28 Mon Sep 17 00:00:00 2001
From: Joshua Bemenderfer <josh@thederf.com>
Date: Sun, 19 Feb 2023 17:04:36 -0500
Subject: [PATCH] Implement basic document functions for C API, mostly
 equivalent to JS ones.

---
 packages/c/.gitignore |   2 +
 packages/c/document.h | 212 ++++++++++++++++++++++++++++++++++++++++++
 packages/c/parser.h   |  10 +-
 3 files changed, 218 insertions(+), 6 deletions(-)
 create mode 100644 packages/c/document.h

diff --git a/packages/c/.gitignore b/packages/c/.gitignore
index a65dd00..a8a24c4 100644
--- a/packages/c/.gitignore
+++ b/packages/c/.gitignore
@@ -1 +1,3 @@
 test/test-runner
+test/document
+test/document.c
\ No newline at end of file
diff --git a/packages/c/document.h b/packages/c/document.h
new file mode 100644
index 0000000..6e6998c
--- /dev/null
+++ b/packages/c/document.h
@@ -0,0 +1,212 @@
+#ifndef TERRACE_DOCUMENT_H
+#define TERRACE_DOCUMENT_H
+
+#include "parser.h"
+
+typedef struct terrace_document_s {
+  // == Internal State == //
+  unsigned int _repeatCurrentLine;
+  // Current line being read
+  char* _currentLine;
+
+  // == External Information == //
+  // Embedded line data struct. Holds information about the current parsed line
+  terrace_linedata_t lineData;
+  // Custom data passed to the readline function
+  void* userData;
+  /**
+   * Line reader function, provided by the user
+   * Needed to get the next line inside of `terrace_next(doc)`
+   * @param {char**} line First argument is a pointer to `_currentLine`, above
+   * @param {void*} userData Second argument is `userData`, above
+   * @returns {int} The number of characters read, or -1 if no characters were read.
+   */
+  int (*reader)(char** line, void* userData);
+} terrace_document_t;
+
+/**
+ * Initialize a Terrace document with indent parameters and the function neded to read lines.
+ * @param {char} indent The indent character to use. Generally a single space character.
+ * @param {int (*reader)(char** line, void* userData)} A function pointer to a function that reads lines sequentially
+ * from a user-provided source. Receives a pointer to lineData->_currLine, and userData, supplied in the next argument.
+ * @param {void*} userData A user-supplied pointer to any state information needed by their reader function.
+ * Passed to `reader`each time it is called.
+ * @returns {terrace_document_t} An initialized document that can now be used for futher parsing.
+ */
+terrace_document_t terrace_create_document(const char indent, int (*reader)(char** line, void* userData), void* userData) {
+  terrace_document_t document = {
+    ._repeatCurrentLine = 0,
+    ._currentLine = 0,
+    .lineData = terrace_create_line_data(indent),
+    .reader = reader,
+    .userData = userData
+  };
+
+  return document;
+}
+
+/**
+ * Returns the number of indent characters of the current line
+ *
+ * Given the following document, `terrace_level(doc)` would return 0, 1, 2, and 5 respectively for each line
+ *
+ * ```terrace
+ * block
+ *  block
+ *   block
+ *      block
+ * ```
+ * @returns {unsigned int} The indent level of the current line
+ */
+unsigned int terrace_level(terrace_document_t* doc) {
+  return doc->lineData.level;
+}
+
+/**
+ * Get a string with the current line contents
+ * If `startOffset` is -1, skips all indent characters by default. Otherwise only skips the amount specified
+ *
+ * Given the following document
+ *
+ * ```terrace
+ * root
+ *     sub-line
+ * ```
+ * `terrace_line(doc, -1)` on the second line returns "sub-line", trimming off the leading indent characters
+ * `terrace_line(doc, 0)` however, returns "    sub-line", with all four leading spaces
+ *
+ * `startOffset`s other than `-1` are primarily used for parsing blocks that have literal indented multi-line text
+ *
+ * @param {terrace_document_t*} doc A pointer to the Terrace document being parsed
+ * @param {int} startOffset How many indent characters to skip before outputting the line contents. Defaults to the current indent level
+ * @returns {char*} The line contents starting from `startOffset`
+ */
+char* terrace_line(terrace_document_t* doc, int startOffset) {
+  if (startOffset == -1) startOffset = doc->lineData.level;
+  return doc->_currentLine + startOffset;
+}
+
+/**
+ * Get the *length* of the first "word" of a line,
+ * starting from the first non-indent character to the first space or end of the line
+ * Often used for deciding how to parse a block.
+ *
+ * Because C uses NULL-terminated strings, we cannot easily slice a string to return something out of the middle.
+ * Instead, `terrace_head_length` provides the length of the head portion.
+ * In combination with `doc->lineData.offsetHead`, you can copy the head section into a new string,
+ * or use any number of `strn*` C stdlib functions to work with the head section without copying it.
+ *
+ * Terrace DSLs do not *need* to use head-tail line structure, but support for them is built into the parser
+ *
+ * Given the following line, `terrace_head_length(doc)` returns `5`
+ *
+ * ```terrace
+ *   title An Important Document
+ * ```
+ * @param {terrace_document_t*} doc A pointer to the Terrace document being parsed
+ * @returns {int} The length of the `head` portion (first word) of a line
+ */
+int terrace_head_length(terrace_document_t* doc) {
+  return doc->lineData.offsetTail - doc->lineData.offsetHead;
+}
+
+/**
+ * Get a char pointer to everything following the first "word" of a line,
+ * starting from the first character after the space at the end of `head`
+ *
+ * Terrace DSLs do not *need* to use head-tail line structure, but support for them is built into the parser
+ *
+ * Given the following line, `terrace_tail(doc)` returns "An Important Document"
+ *
+ * ```terrace
+ *   title An Important Document
+ * ```
+ * @param {terrace_document_t*} doc A pointer to the Terrace document being parsed
+ * @returns {char*} The remainder of the line following the `head` portion, with no leading space
+ */
+char* terrace_tail(terrace_document_t* doc) {
+  return doc->_currentLine + doc->lineData.offsetTail + 1;
+}
+
+/**
+ * Quickly check if the current line head matches a specified value. Useful in many document-parsing situations.
+ *
+ * Given the following line
+ *
+ * ```terrace
+ * title An Important Document
+ * ```
+ *
+ * `terrace_match(doc, "title")` returns `1`
+ * `terrace_match(doc, "somethingElse") returns `0`
+ *
+ * @param {const char*} matchValue A string to check against the line `head` for equality
+ * @returns {char} A byte set to 0 if the head does not match, or 1 if it does match
+ */
+char terrace_match(terrace_document_t* doc, const char* matchHead) {
+  // Get a pointer to the start of the head portion of the string.
+  char* head = doc->_currentLine + doc->lineData.offsetHead;
+
+  int i = 0;
+  // Loop until we run out of characters in `matchHead`.
+  while (matchHead[i] != '\0') {
+    // Return as unmatched if we run out of `head` characters
+    // or if a character at the same position in both matchHead and head is not identical.
+    if (head[i] == '\0' || matchHead[i] != head[i]) return 0;
+    i++;
+  }
+
+  // If we didn't return inside the while loop, `matchHead` and `head` are equivalent, a successful match.
+  return 1;
+}
+
+/**
+  * Advances the current position in the terrace document and populates `doc->lineData`
+  * with the parsed information from that line
+  *
+  * Returns `1` after parsing the next line, or `0` upon reaching the end of the document.
+  * If the `levelScope` parameter is not -1, `terrace_next()` will also return `0` when it encounters a line
+  * with a level at or below `levelScope`. This allows you to iterate through subsections of a document.
+  *
+  * If a lower-level line was encountered, the following call to `next()` will repeat this line again.
+  * This allows a child loop to look forward, determine that the next line will be outside its purview,
+  * and return control to the calling loop transparently without additional logic.
+  *
+  * Intended to be used inside a while loop to parse a section of a Terrace document.
+  *
+  * ```c
+  * while(terrace_next(doc, -1)) {
+  *   // Do something with each line.
+  * }
+  * ```
+  *
+  * @param {number} levelScope If set above -1, `next()` will return `0` when it encounters a line with a level at or below `levelScope`
+  * @returns {char} Returns `1` after parsing a line, or `0` if the document has ended or a line at or below `levelScope` has been encountered.
+  */
+char terrace_next(terrace_document_t* doc, int levelScope) {
+  // Repeat the current line instead of parsing a new one if the previous call to next()
+  // determined the current line to be out of its scope.
+  if (doc->_repeatCurrentLine) doc->_repeatCurrentLine = 0;
+  // Otherwise parse the line normally.
+  else {
+    // Load the next line from the line reader.
+    int chars_read = doc->reader(&doc->_currentLine, doc->userData);
+    // If there are no more lines, bail out.
+    if (chars_read == -1) return 0;
+
+    // Populate lineData with parsed information from the current line.
+    terrace_parse_line(doc->_currentLine, &doc->lineData);
+  }
+
+  // If we shouldn't be handling this line, make the following call to next() repeat the current line.
+  // Allows a child loop to look forward, determine that the next line will be outside its purview,
+  // and return control to the calling loop transparently without additional logic.
+  if ((int) terrace_level(doc) <= levelScope) {
+    doc->_repeatCurrentLine = 1;
+    return 0;
+  }
+
+  return 1;
+}
+
+#endif
\ No newline at end of file
diff --git a/packages/c/parser.h b/packages/c/parser.h
index 5bd5aea..1a86ad8 100644
--- a/packages/c/parser.h
+++ b/packages/c/parser.h
@@ -2,7 +2,7 @@
 #define TERRACE_PARSER_H
 
 // Holds the parsed information from each line.
-struct terrace_linedata_s {
+typedef struct terrace_linedata_s {
   // Which character is being used for indentation. Avoids having to specify it on each terrace_parse_line call.
   char indent;
   // How many indent characters are present in the current line before the first non-indent character.
@@ -12,11 +12,9 @@ struct terrace_linedata_s {
   unsigned int offsetHead;
   // The number of characters before the start of the line's "tail" section.
   unsigned int offsetTail;
-};
+} terrace_linedata_t;
 
-typedef struct terrace_linedata_s terrace_linedata_t;
-
-terrace_linedata_t terrace_create_line_data(char indent) {
+terrace_linedata_t terrace_create_line_data(const char indent) {
   terrace_linedata_t line_data = { .indent = indent, .level = 0, .offsetHead = 0, .offsetTail = 0 };
   return line_data;
 }
@@ -26,7 +24,7 @@ terrace_linedata_t terrace_create_line_data(char indent) {
  * @param char* line A pointer to the line to parse as a C-style string. Shouldn't end with a newline.
  * @param terrace_linedata_t* lineData A pointer to the terrace_linedata_t struct to store information about the current line in.
  */
-void terrace_parse_line(char *line, terrace_linedata_t *lineData) {
+void terrace_parse_line(const char *line, terrace_linedata_t *lineData) {
   // Empty lines are nullptr/0 as they have no characters. (The newline character should be stripped off.)
   // Special case handling for these allows them to be parsed extra quickly.
   if (!line) {