Syntax highlighting of one line of text (text_hl.h): the rules of a bench text, JSON and CSV. More...

#include "text_hl.h"
#include <stdbool.h>
#include <string.h>

Data Structures
struct	spans_t
	Where the spans of a line are written. More...

Macros
#define	CHECKSUM_DIGITS 8u
	Digits of the checksum in a KST file's "checksum" line.
#define	HEADING_MIN 3u
	Characters a line has at least to be a heading.
#define	CLASS_CSS(name, css, role)
	Expands a row of TEXT_HL_CLASS_ROWS to its style sheet name.

Enumerations
enum	letter_t { LETTER_NONE = 0 , LETTER_UPPER , LETTER_LOWER }
	The case of a letter. More...

Functions
static bool	is_word (unsigned char b)
	Whether `b` belongs to a word: an ASCII letter or digit, "_", or any byte of a UTF-8 sequence.
static bool	is_digit (unsigned char b)
	Whether `b` is an ASCII digit.
static bool	is_hex (unsigned char b)
	Whether `b` is a hexadecimal digit.
static bool	is_blank (unsigned char b)
	Whether `b` is a space or a tab.
static letter_t	letter (const unsigned char s, size_t n, size_t len)
	The case of the letter at `s:` ASCII and the 7 German letters.
static void	emit (spans_t *sp, size_t start, size_t len, text_hl_class_t cls)
	Add bytes `start` to `start` + `len` as class `cls`, joined to the last span when it has that class.
static void	trim (const char s, size_t a, size_t *b)
	Narrow [a, b) of `s` past the blanks at both ends.
const char *	text_hl_word_core (const char s, size_t len)
	The word a writer's string gives: spaces taken off both ends, a "Label:" and the spaces after it taken off the front, then one pair of parentheses around it and a full stop behind it.
static bool	apart (unsigned char e, unsigned char nb)
	Whether the edge byte `e` of a word stands alone against its neighbour `nb:` a word byte against no word byte, a mark against anything but the same mark.
static size_t	word_at (const unsigned char line, size_t len, size_t i, const char const *list, size_t n)
	The longest verdict word of `list` at byte `i` of the line.
text_hl_class_t	text_hl_unit_class (const char *unit, size_t len)
	The class of a unit of the table.
const char *	text_hl_class_css (text_hl_class_t cls)
	The style sheet's name of a class (TEXT_HL_CLASS_ROWS).
static size_t	unit_end (const unsigned char *line, size_t len, size_t k)
	The end of a unit's bytes from byte `k:` a "%"; or a "/" and letters, or letters; the ohm sign counts as a letter.
static text_hl_class_t	unit_class_at (const unsigned char *line, size_t len, size_t k, size_t u)
	The class of the unit from byte `k` to `u`, when it ends at no word byte.
static size_t	number_at (const unsigned char line, size_t len, size_t i, text_hl_class_t cls)
	A number at byte `i`, with the unit behind it: joined to it, as in "6.00V", or after one space, as the older reports write it.
static bool	hex_line (const unsigned char *s, size_t len)
	Whether a line is a register line "NN: XX XX ...", a checksum line "checksum XXXXXXXX" or a journal line "key=hexdigits", trailing spaces allowed.
static bool	heading (const unsigned char *s, size_t len)
	Whether a line is a heading: its first character an upper-case letter, at least HEADING_MIN characters, and no lower-case letter before the first "(".
static size_t	label_end (const unsigned char *s, size_t len)
	Where a label ends: after the first ":" or "=" in the first TEXT_HL_LABEL_CELLS characters, when a blank or the line's end follows it and the line starts with no blank.
static void	rest (spans_t sp, const unsigned char s, size_t len, size_t i, const text_hl_words_t *words)
	The spans of a bench text's line from byte `i` on: comments, verdict words and numbers with units.
static void	text_line (spans_t sp, const unsigned char s, size_t len, size_t line_no, const text_hl_words_t *words)
	The spans of a bench text's line.
static size_t	json_number (const unsigned char *s, size_t len, size_t i)
	Bytes of a JSON number at `i`.
static void	json_line (spans_t sp, const unsigned char s, size_t len)
	The spans of a JSON line.
static void	csv_line (spans_t sp, const unsigned char s, size_t len, size_t line_no)
	The spans of a CSV line.
size_t	text_hl_line (const char line, size_t len, size_t line_no, text_hl_kind_t kind, const text_hl_words_t words, text_hl_span_t *out, size_t cap)
	The spans of one line.
static bool	same_ext (const char s, size_t n, const char ext)
	Whether the `n` bytes at `s` are `ext` without regard to case.
static size_t	ext_at (const char *name, size_t n)
	Where the extension of the first `n` bytes of `name` starts.
text_hl_kind_t	text_hl_kind_of (const char *name)
	How a file is classed, by its name: ".json" JSON, ".csv" CSV, ".txt" TEXT, in any case, a ".NEW" behind any of them taken off first; PLAIN for any other name.

Variables
const text_hl_unit_t	text_hl_units [TEXT_HL_UNIT_COUNT]
	The units a number may carry, case as written: V mV A mA W rpm Hz us ms s /s deg Grad % mOhm B kB MB mΩ Ω; mOhm for the reports of 0.22 and older. test_text_hl holds it to every unit the report formats of ui_text.def put behind a number.
static const char *const	k_css [TEXT_HL_CLASSES]
	The style sheet's name of each class, by class.

Detailed Description

Syntax highlighting of one line of text (text_hl.h): the rules of a bench text, JSON and CSV.

SPDX-License-Identifier: MIT

Macro Definition Documentation

◆ CLASS_CSS

#define CLASS_CSS	(	name,
		css,
		role )

Value:

css,

Expands a row of TEXT_HL_CLASS_ROWS to its style sheet name.

Enumeration Type Documentation

◆ letter_t

enum letter_t

The case of a letter.

Enumerator
LETTER_NONE	no letter
LETTER_UPPER	A to Z, Ä Ö Ü
LETTER_LOWER	a to z, ä ö ü ß

Function Documentation

◆ apart()

bool apart	(	unsigned char	e,
		unsigned char	nb )

static

Whether the edge byte e of a word stands alone against its neighbour nb: a word byte against no word byte, a mark against anything but the same mark.

Parameters

e	The word's first or last byte.
nb	The byte beside it in the line.

Returns: true when the word ends there.

◆ csv_line()

void csv_line	(	spans_t *	sp,
		const unsigned char *	s,
		size_t	len,
		size_t	line_no )

static

The spans of a CSV line.

Parameters

sp	The spans.
s	The line.
len	Its bytes.
line_no	Its index in the file: 0 is the header.

◆ emit()

void emit	(	spans_t *	sp,
		size_t	start,
		size_t	len,
		text_hl_class_t	cls )

static

Add bytes start to start + len as class cls, joined to the last span when it has that class.

Parameters

sp	The spans.
start	First byte; the end of the last span.
len	Bytes; 0 adds nothing.
cls	The class.

◆ ext_at()

size_t ext_at	(	const char *	name,
		size_t	n )

static

Where the extension of the first n bytes of name starts.

Parameters

name	The name.
n	Bytes to look at.

Returns: The offset of its last ".", or n without one.

◆ heading()

bool heading	(	const unsigned char *	s,
		size_t	len )

static

Whether a line is a heading: its first character an upper-case letter, at least HEADING_MIN characters, and no lower-case letter before the first "(".

Parameters

s	The line.
len	Its bytes.

Returns: true for a heading.

◆ hex_line()

bool hex_line	(	const unsigned char *	s,
		size_t	len )

static

Whether a line is a register line "NN: XX XX ...", a checksum line "checksum XXXXXXXX" or a journal line "key=hexdigits", trailing spaces allowed.

Parameters

s	The line.
len	Its bytes.

Returns: true for a hex line.

◆ is_blank()

bool is_blank ( unsigned char b )

static

Whether b is a space or a tab.

Parameters

b The byte.

Returns: true for 0x20 and 0x09.

◆ is_digit()

bool is_digit ( unsigned char b )

static

Whether b is an ASCII digit.

Parameters

b The byte.

Returns: true for 0 to 9.

◆ is_hex()

bool is_hex ( unsigned char b )

static

Whether b is a hexadecimal digit.

Parameters

b The byte.

Returns: true for 0 to 9, A to F and a to f.

◆ is_word()

bool is_word ( unsigned char b )

static

Whether b belongs to a word: an ASCII letter or digit, "_", or any byte of a UTF-8 sequence.

Parameters

b The byte.

Returns: true for a word byte.

◆ json_line()

void json_line	(	spans_t *	sp,
		const unsigned char *	s,
		size_t	len )

static

The spans of a JSON line.

Parameters

sp	The spans.
s	The line.
len	Its bytes.

◆ json_number()

size_t json_number	(	const unsigned char *	s,
		size_t	len,
		size_t	i )

static

Bytes of a JSON number at i.

Parameters

s	The line.
len	Its bytes.
i	Where it would start.

Returns: Bytes; 0 when no number starts here.

◆ label_end()

size_t label_end	(	const unsigned char *	s,
		size_t	len )

static

Where a label ends: after the first ":" or "=" in the first TEXT_HL_LABEL_CELLS characters, when a blank or the line's end follows it and the line starts with no blank.

Parameters

s	The line.
len	Its bytes.

Returns: Bytes of the label with its mark; 0 for none.

◆ letter()

letter_t letter	(	const unsigned char *	s,
		size_t	n,
		size_t *	len )

static

The case of the letter at s: ASCII and the 7 German letters.

Parameters

s	The bytes.
n	Their count from `s`, at least 1.
len	Out: bytes of the character, 1 or 2.

Returns: The case; LETTER_NONE for any other character.

◆ number_at()

size_t number_at	(	const unsigned char *	line,
		size_t	len,
		size_t	i,
		text_hl_class_t *	cls )

static

A number at byte i, with the unit behind it: joined to it, as in "6.00V", or after one space, as the older reports write it.

A word joined to the number that is no unit of the table ("6.00Volt", "5x") makes the whole no number.

Parameters

line	The line.
len	Its bytes.
i	Where the number would start.
cls	Out: the unit's class; TEXT_HL_TEXT for a bare number.

Returns: Bytes of the number and its unit; 0 when no number starts here.

◆ rest()

void rest	(	spans_t *	sp,
		const unsigned char *	s,
		size_t	len,
		size_t	i,
		const text_hl_words_t *	words )

static

The spans of a bench text's line from byte i on: comments, verdict words and numbers with units.

Parameters

sp	The spans.
s	The line.
len	Its bytes.
i	Where to start.
words	The verdict words; NULL for none.

◆ same_ext()

bool same_ext	(	const char *	s,
		size_t	n,
		const char *	ext )

static

Whether the n bytes at s are ext without regard to case.

Parameters

s	The bytes.
n	Their count.
ext	An extension in lower case, with its dot.

Returns: true when they are the same.

◆ text_hl_class_css()

const char * text_hl_class_css ( text_hl_class_t cls )

The style sheet's name of a class (TEXT_HL_CLASS_ROWS).

Parameters

cls	The class.

Returns: Its short name; "tx" for a value out of range.

◆ text_hl_kind_of()

text_hl_kind_t text_hl_kind_of ( const char * name )

How a file is classed, by its name: ".json" JSON, ".csv" CSV, ".txt" TEXT, in any case, a ".NEW" behind any of them taken off first; PLAIN for any other name.

Parameters

name	The file name; NULL is PLAIN.

Returns: The kind.

◆ text_hl_line()

size_t text_hl_line	(	const char *	line,
		size_t	len,
		size_t	line_no,
		text_hl_kind_t	kind,
		const text_hl_words_t *	words,
		text_hl_span_t *	out,
		size_t	cap )

The spans of one line.

With room for fewer spans than the line has, the last span the room holds is TEXT and reaches to the line's end.

Parameters

line	The line's bytes, without its line end.
len	Their count.
line_no	The line's index in the file, from 0.
kind	How the file is classed.
words	The verdict words; NULL for none.
out	Out: the spans.
cap	Room in `out`.

Returns: Spans written; 0 for an empty line, a NULL or cap 0.

◆ text_hl_unit_class()

text_hl_class_t text_hl_unit_class	(	const char *	unit,
		size_t	len )

The class of a unit of the table.

Parameters

unit	The unit's bytes.
len	Their count.

Returns: TEXT_HL_VOLT, TEXT_HL_CURR, TEXT_HL_POWER, TEXT_HL_RATE or TEXT_HL_UNIT; TEXT_HL_TEXT for a word that is no unit.

◆ text_hl_word_core()

const char * text_hl_word_core	(	const char *	s,
		size_t *	len )

The word a writer's string gives: spaces taken off both ends, a "Label:" and the spaces after it taken off the front, then one pair of parentheses around it and a full stop behind it.

"Brown-out: not run" gives "not run", " (cut short)" "cut short", "Not run." "Not run".

Parameters

s	The string; NULL gives an empty word.
len	Out: bytes of the word.

Returns: Its first byte, inside s.

◆ text_line()

void text_line	(	spans_t *	sp,
		const unsigned char *	s,
		size_t	len,
		size_t	line_no,
		const text_hl_words_t *	words )

static

The spans of a bench text's line.

Parameters

sp	The spans.
s	The line.
len	Its bytes.
line_no	Its index in the file.
words	The verdict words; NULL for none.

◆ trim()

void trim	(	const char *	s,
		size_t *	a,
		size_t *	b )

static

Narrow [*a, *b) of s past the blanks at both ends.

Parameters

s	The string.
a	In and out: the first byte.
b	In and out: the byte after the last.

◆ unit_class_at()

text_hl_class_t unit_class_at	(	const unsigned char *	line,
		size_t	len,
		size_t	k,
		size_t	u )

static

The class of the unit from byte k to u, when it ends at no word byte.

Parameters

line	The line.
len	Its bytes.
k	The unit's first byte.
u	The byte after its last (unit_end()).

Returns: The unit's class; TEXT_HL_TEXT for none, a word that is no unit, or one a word byte runs on from.

◆ unit_end()

size_t unit_end	(	const unsigned char *	line,
		size_t	len,
		size_t	k )

static

The end of a unit's bytes from byte k: a "%"; or a "/" and letters, or letters; the ohm sign counts as a letter.

Parameters

line	The line.
len	Its bytes.
k	Where the unit would start.

Returns: The byte after the unit's last; k for none.

◆ word_at()

size_t word_at	(	const unsigned char *	line,
		size_t	len,
		size_t	i,
		const char const	list,
		size_t	n )

static

The longest verdict word of list at byte i of the line.

Parameters

line	The line.
len	Its bytes.
i	Where the word would start.
list	The words.
n	Their count.

Returns: Bytes of the longest word that matches; 0 for none.

Variable Documentation

◆ k_css

const char* const k_css[TEXT_HL_CLASSES]

static

Initial value:

                              = {
 
}

The style sheet's name of each class, by class.

◆ text_hl_units

const text_hl_unit_t text_hl_units[TEXT_HL_UNIT_COUNT]

Initial value:

                                   = {
    {"V", TEXT_HL_VOLT},     {"mV", TEXT_HL_VOLT},  {"A", TEXT_HL_CURR},
    {"mA", TEXT_HL_CURR},    {"W", TEXT_HL_POWER},  {"rpm", TEXT_HL_RATE},
    {"Hz", TEXT_HL_RATE},    {"us", TEXT_HL_UNIT},  {"ms", TEXT_HL_UNIT},
    {"s", TEXT_HL_UNIT},     {"/s", TEXT_HL_UNIT},  {"deg", TEXT_HL_UNIT},
    {"Grad", TEXT_HL_UNIT},  {"%", TEXT_HL_UNIT},   {"mOhm", TEXT_HL_UNIT},
    {"B", TEXT_HL_UNIT},     {"kB", TEXT_HL_UNIT},  {"MB", TEXT_HL_UNIT},
    {"mΩ", TEXT_HL_UNIT},    {"Ω", TEXT_HL_UNIT},
}

The units a number may carry, case as written: V mV A mA W rpm Hz us ms s /s deg Grad % mOhm B kB MB mΩ Ω; mOhm for the reports of 0.22 and older. test_text_hl holds it to every unit the report formats of ui_text.def put behind a number.

Data Structures

Macros

Enumerations

Functions

Variables

Detailed Description

Macro Definition Documentation

◆ CLASS_CSS

Enumeration Type Documentation

◆ letter_t

Function Documentation

◆ apart()

◆ csv_line()

◆ emit()

◆ ext_at()

◆ heading()

◆ hex_line()

◆ is_blank()

◆ is_digit()

◆ is_hex()

◆ is_word()

◆ json_line()

◆ json_number()

◆ label_end()

◆ letter()

◆ number_at()

◆ rest()

◆ same_ext()

◆ text_hl_class_css()

◆ text_hl_kind_of()

◆ text_hl_line()

◆ text_hl_unit_class()

◆ text_hl_word_core()

◆ text_line()

◆ trim()

◆ unit_class_at()

◆ unit_end()

◆ word_at()

Variable Documentation

◆ k_css

◆ text_hl_units