///|
/// Return a human-readable simple-font subtype label.
///
/// Type 3 labels include the declared CharProc names to make debugging parsed
/// glyph programs easier.
pub fn PdfSimpleFontType::debug_name(self : PdfSimpleFontType) -> String {
match self {
PdfType1 => "Type 1"
PdfMMType1 => "MM Type 1"
PdfType3(data) => {
let mut text = "Type 3: charprocs for"
for entry in data.charprocs {
let (name, _) = entry
text = text + pdf_text_string_of_name(name)
}
text
}
PdfTrueType => "Truetype"
}
}
///|
/// Return a multi-line debug summary for this simple font.
pub fn PdfSimpleFont::debug_name(self : PdfSimpleFont) -> String {
let basefont = match self.basefont {
Some(name) => pdf_text_string_of_name(name)
None => ""
}
"fonttype = " +
self.fonttype.debug_name() +
"\nbasefont " +
basefont +
"\nencoding is " +
self.encoding.debug_name() +
"\n"
}
///|
/// One `/Differences` entry in a simple-font encoding dictionary.
///
/// `code` is the one-byte character code that maps to `name`.
pub(all) struct PdfEncodingDifference {
name : PdfName
code : Int
} derive(Debug, Eq, ToJson)
///|
/// Simple-font character encoding.
///
/// `PdfCustomEncoding` layers `/Differences` over a base encoding.
/// `PdfFillUndefinedWithStandard` models TrueType behavior where undefined
/// entries fall back to StandardEncoding.
pub(all) enum PdfEncoding {
PdfImplicitInFontFile
PdfStandardEncoding
PdfMacRomanEncoding
PdfWinAnsiEncoding
PdfMacExpertEncoding
PdfCustomEncoding(PdfEncoding, Array[PdfEncodingDifference])
PdfFillUndefinedWithStandard(PdfEncoding)
} derive(Debug, Eq, ToJson)
///|
/// Parsed font model used by text extraction and font writing.
pub(all) enum PdfFont {
PdfFontStandard(PdfStandardFontData)
PdfFontSimple(PdfSimpleFont)
PdfFontCIDKeyed(PdfCIDKeyedFont)
} derive(Debug, Eq, ToJson)
///|
/// Text extractor bound to a parsed font.
///
/// The extractor converts PDF text-string bytes to character codes, glyph
/// names, and Unicode codepoints according to the font's encoding data.
pub(all) struct PdfTextExtractor {
font : PdfFont
} derive(Debug, Eq, ToJson)
///|
/// Extracted glyph result for one character code.
///
/// `glyphname` is the resolved PDF glyph name and `codepoints` is the Unicode
/// scalar sequence inferred from ToUnicode, CMap, or Adobe glyph-list data.
pub(all) struct PdfTextGlyph {
glyphname : PdfName
codepoints : Array[Int]
} derive(Debug, Eq, ToJson)
///|
/// Return a human-readable label for a parsed font.
pub fn PdfFont::debug_name(self : PdfFont) -> String {
match self {
PdfFontStandard(font) => font.debug_name()
PdfFontSimple(font) => "SimpleFont " + font.debug_name()
PdfFontCIDKeyed(font) => font.debug_name()
}
}
///|
/// Compatibility wrapper for `PdfFont::debug_name`.
pub fn pdf_string_of_font(font : PdfFont) -> String {
font.debug_name()
}
///|
/// Test whether this font is a Type 0 font using `/Identity-H`.
pub fn PdfFont::is_identity_h(self : PdfFont) -> Bool {
match self {
PdfFontCIDKeyed(font) => font.is_identity_h()
_ => false
}
}
///|
/// Compatibility wrapper for `PdfFont::is_identity_h`.
pub fn pdf_is_identity_h(font : PdfFont) -> Bool {
font.is_identity_h()
}
///|
/// Test whether this font is a Type 0 font using `/Identity-V`.
pub fn PdfFont::is_identity_v(self : PdfFont) -> Bool {
match self {
PdfFontCIDKeyed(font) => font.is_identity_v()
_ => false
}
}
///|
/// Return true when this font consumes fixed two-byte text character codes.
pub fn PdfFont::uses_two_byte_codes(self : PdfFont) -> Bool {
match self {
PdfFontCIDKeyed(font) => font.uses_two_byte_codes()
_ => false
}
}
///|
/// Build a text extractor for an already parsed font.
pub fn PdfFont::text_extractor(self : PdfFont) -> PdfTextExtractor {
{ font: self }
}
///|
/// Compatibility wrapper for `PdfFont::text_extractor`.
pub fn pdf_text_extractor_of_font_real(font : PdfFont) -> PdfTextExtractor {
font.text_extractor()
}
///|
/// Build a pure reverse lookup closure for an already parsed font.
///
/// The returned closure maps one Unicode codepoint to the first matching PDF
/// character code when the font can represent it. `debug` is accepted for
/// CamlPDF API compatibility; this implementation does not write diagnostics.
pub fn pdf_charcode_extractor_of_font_real(
font : PdfFont,
debug? : Bool = false,
) -> (Int) -> Int? {
// CamlPDF reports missing glyphs to stderr when debug is true; this wrapper
// keeps the lookup pure while preserving the API shape.
ignore(debug)
let extractor = font.text_extractor()
fn(codepoint) { extractor.charcode_of_codepoint(codepoint) }
}
///|
fn pdf_text_width_of_standard_charcode(
font : PdfStandardFont,
encoding : PdfEncoding,
code : Int,
) -> Double {
if code >= 0 && code <= 255 {
let bytes : FixedArray[Byte] = [code.to_byte()]
font.text_width(encoding, Bytes::from_array(bytes)).to_double()
} else {
0.0
}
}
///|
fn pdf_text_simple_font_charcode_width(
font : PdfSimpleFont,
code : Int,
) -> Double {
match font.fontmetrics {
Some(metrics) =>
if code >= 0 && code < metrics.length() {
metrics[code]
} else {
0.0
}
None =>
match font.basefont {
Some(basefont) =>
match pdf_standard_font_of_pdf_name(basefont) {
Some(standard_font) =>
pdf_text_width_of_standard_charcode(
standard_font,
font.encoding,
code,
)
None => 0.0
}
None => 0.0
}
}
}
///|
fn pdf_text_cid_font_charcode_width(
font : PdfCIDKeyedFont,
code : Int,
) -> Double {
for entry in font.composite.cid_widths {
if entry.0 == code {
break entry.1
}
} nobreak {
font.composite.cid_default_width
}
}
///|
/// Measure PDF text bytes in glyph-space units for this extractor's font.
///
/// Text bytes are segmented according to the font encoding. Malformed byte
/// sequences raise `PdfError`, and missing metrics contribute width `0`.
pub fn PdfTextExtractor::text_width(
self : PdfTextExtractor,
text : BytesView,
) -> Double raise PdfError {
match self.font {
PdfFontStandard(font) => font.text_width(text).to_double()
PdfFontSimple(font) => {
let mut width = 0.0
for code in self.font.pdf_text_charcodes_of_text(text) {
width += pdf_text_simple_font_charcode_width(font, code)
}
width
}
PdfFontCIDKeyed(font) => {
let mut width = 0.0
for code in self.font.pdf_text_charcodes_of_text(text) {
width += pdf_text_cid_font_charcode_width(font, code)
}
width
}
}
}
///|
/// Return a human-readable encoding label.
pub fn PdfEncoding::debug_name(self : PdfEncoding) -> String {
match self {
PdfImplicitInFontFile => "ImplicitInFontFile"
PdfStandardEncoding => "StandardEncoding"
PdfMacRomanEncoding => "MacRomanEncoding"
PdfWinAnsiEncoding => "WinAnsiEncoding"
PdfMacExpertEncoding => "MacExpertEncoding"
PdfCustomEncoding(base, differences) => {
let mut text = "CustomEncoding with base " +
base.debug_name() +
" and differences "
for difference in differences {
text = text +
difference.code.to_string() +
" -> " +
pdf_text_string_of_name(difference.name) +
", "
}
text
}
PdfFillUndefinedWithStandard(base) =>
"FillUndefinedWithStandard " + base.debug_name()
}
}
///|
/// Return the PDF name for a built-in simple-font encoding.
///
/// Custom, implicit, and fallback encodings do not have a direct PDF name and
/// return `None`.
pub fn PdfEncoding::pdf_name(self : PdfEncoding) -> PdfName? {
match self {
PdfStandardEncoding => Some(pdf_text_standard_encoding_name)
PdfMacRomanEncoding => Some(pdf_text_mac_roman_encoding_name)
PdfWinAnsiEncoding => Some(pdf_text_win_ansi_encoding_name)
PdfMacExpertEncoding => Some(pdf_text_mac_expert_encoding_name)
_ => None
}
}
///|
/// Recognize a built-in simple-font encoding name.
pub fn pdf_encoding_of_name(name : PdfName) -> PdfEncoding? {
match name {
name if name == pdf_text_standard_encoding_name => Some(PdfStandardEncoding)
name if name == pdf_text_mac_roman_encoding_name =>
Some(PdfMacRomanEncoding)
name if name == pdf_text_win_ansi_encoding_name => Some(PdfWinAnsiEncoding)
name if name == pdf_text_mac_expert_encoding_name =>
Some(PdfMacExpertEncoding)
_ => None
}
}