minori: src/core/strings.cc annotate

annotate src/core/strings.cc @ 321:8141f409d52c

services/anilist: fix getting producers should studios and producers be stored separately?

author	Paper <paper@paper.us.eu.org>
date	Wed, 12 Jun 2024 20:42:44 -0400
parents	1b5c04268d6a
children	c32467cd06bb

rev	line source
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	1 /**
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	2 * strings.cpp: Useful functions for manipulating strings
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	3 **/
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	4 #include "core/strings.h"
187 9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	5 #include "core/session.h" // locale
9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	6
64 fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	7 #include <QByteArray>
101 c537996cf67b : multitude of config changes Paper <mrpapersonic@gmail.com>* parents: 100 diff changeset	8 #include <QDebug>
258 862d0d8619f6 : HUUUGE changes Paper <paper@paper.us.eu.org>* parents: 250 diff changeset	9 #include <QLocale>
64 fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	10 #include <QString>
274 f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	11 #include <QTextDocument>
187 9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	12
15 cde8f67a7c7d : update, megacommit :) Paper <mrpapersonic@gmail.com>* parents: 9 diff changeset	13 #include <algorithm>
cde8f67a7c7d : update, megacommit :) Paper <mrpapersonic@gmail.com>* parents: 9 diff changeset	14 #include <cctype>
62 4c6dd5999b39 : update Paper <mrpapersonic@gmail.com>* parents: 15 diff changeset	15 #include <codecvt>
258 862d0d8619f6 : HUUUGE changes Paper <paper@paper.us.eu.org>* parents: 250 diff changeset	16 #include <iostream>
273 f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	17 #include <iomanip>
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	18 #include <locale>
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	19 #include <string>
258 862d0d8619f6 : HUUUGE changes Paper <paper@paper.us.eu.org>* parents: 250 diff changeset	20 #include <unordered_map>
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	21 #include <vector>
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	22
264 9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	23 #include "utf8proc.h"
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	24
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	25 namespace Strings {
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	26
99 503bc1547d49 strings: clarify on some functions and make some of them miniscule Paper <mrpapersonic@gmail.com> parents: 98 diff changeset	27 /* ew */
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	28 std::string Implode(const std::vector<std::string>& vector, const std::string& delimiter) {
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	29 if (vector.size() < 1)
250 c130f47f6f48 : many many changes Paper <paper@paper.us.eu.org>* parents: 231 diff changeset	30 return "";
187 9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	31
221 53211cb1e7f5 library: add initial library stuff Paper <paper@paper.us.eu.org> parents: 211 diff changeset	32 std::string out;
187 9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	33
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	34 for (unsigned long long i = 0; i < vector.size(); i++) {
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	35 out.append(vector.at(i));
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	36 if (i < vector.size() - 1)
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	37 out.append(delimiter);
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	38 }
187 9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	39
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	40 return out;
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	41 }
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	42
258 862d0d8619f6 : HUUUGE changes Paper <paper@paper.us.eu.org>* parents: 250 diff changeset	43 std::vector<std::string> Split(const std::string& text, const std::string& delimiter) {
250 c130f47f6f48 : many many changes Paper <paper@paper.us.eu.org>* parents: 231 diff changeset	44 if (text.length() < 1)
c130f47f6f48 : many many changes Paper <paper@paper.us.eu.org>* parents: 231 diff changeset	45 return {};
c130f47f6f48 : many many changes Paper <paper@paper.us.eu.org>* parents: 231 diff changeset	46
118 39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	47 std::vector<std::string> tokens;
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	48
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	49 std::size_t start = 0, end = 0;
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	50 while ((end = text.find(delimiter, start)) != std::string::npos) {
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	51 tokens.push_back(text.substr(start, end - start));
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	52 start = end + delimiter.length();
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	53 }
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	54 tokens.push_back(text.substr(start));
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	55
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	56 return tokens;
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	57 }
39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	58
99 503bc1547d49 strings: clarify on some functions and make some of them miniscule Paper <mrpapersonic@gmail.com> parents: 98 diff changeset	59 /* This function is really only used for cleaning up the synopsis of
221 53211cb1e7f5 library: add initial library stuff Paper <paper@paper.us.eu.org> parents: 211 diff changeset	60 * horrible HTML debris from AniList :)
258 862d0d8619f6 : HUUUGE changes Paper <paper@paper.us.eu.org>* parents: 250 diff changeset	61 */
260 dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	62 void ReplaceAll(std::string& string, std::string_view find, std::string_view replace) {
98 582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	63 size_t pos = 0;
582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	64 while ((pos = string.find(find, pos)) != std::string::npos) {
582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	65 string.replace(pos, find.length(), replace);
582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	66 pos += replace.length();
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	67 }
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	68 }
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	69
260 dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	70 void SanitizeLineEndings(std::string& string) {
114 ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	71 /* LOL */
260 dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	72 ReplaceAll(string, "\r\n", "\n");
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	73 ReplaceAll(string, "</p>", "\n");
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	74 ReplaceAll(string, "<br>", "\n");
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	75 ReplaceAll(string, "<br />", "\n");
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	76 ReplaceAll(string, "\n\n\n", "\n\n");
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	77 }
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	78
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	79 void ConvertRomanNumerals(std::string& string) {
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	80 static const std::vector<std::pair<std::string_view, std::string_view>> vec = {
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	81 {"2", "II"}, {"3", "III"}, {"4", "IV"}, {"5", "V"}, {"6", "VI"},
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	82 {"7", "VII"}, {"8", "VIII"}, {"9", "IX"}, {"11", "XI"}, {"12", "XII"},
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	83 {"13", "XIII"}
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	84 };
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	85
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	86 for (const auto& item : vec)
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	87 ReplaceAll(string, item.second, item.first);
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	88 }
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	89
264 9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	90 /* this also performs case folding, so our string is lowercase after this */
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	91 void NormalizeUnicode(std::string& string) {
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	92 static constexpr utf8proc_option_t options = static_cast<utf8proc_option_t>(
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	93 UTF8PROC_COMPAT \| UTF8PROC_COMPOSE \| UTF8PROC_STABLE \|
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	94 UTF8PROC_IGNORE \| UTF8PROC_STRIPCC \| UTF8PROC_STRIPMARK \|
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	95 UTF8PROC_LUMP \| UTF8PROC_CASEFOLD \| UTF8PROC_NLF2LS
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	96 );
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	97
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	98 /* ack */
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	99 utf8proc_uint8_t* buf = nullptr;
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	100
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	101 const utf8proc_ssize_t size = utf8proc_map(
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	102 reinterpret_cast<const utf8proc_uint8_t*>(string.data()),
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	103 string.size(),
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	104 &buf,
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	105 options
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	106 );
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	107
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	108 if (size)
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	109 string = std::string(reinterpret_cast<const char*>(buf), size);
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	110
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	111 if (buf)
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	112 free(buf);
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	113 }
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	114
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	115 void NormalizeAnimeTitle(std::string& string) {
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	116 ConvertRomanNumerals(string);
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	117 NormalizeUnicode(string);
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	118 RemoveLeadingChars(string, ' ');
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	119 RemoveTrailingChars(string, ' ');
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	120 }
9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	121
98 582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	122 /* removes dumb HTML tags because anilist is aids and
250 c130f47f6f48 : many many changes Paper <paper@paper.us.eu.org>* parents: 231 diff changeset	123 * gives us HTML for synopses :/
258 862d0d8619f6 : HUUUGE changes Paper <paper@paper.us.eu.org>* parents: 250 diff changeset	124 */
260 dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	125 void RemoveHtmlTags(std::string& string) {
99 503bc1547d49 strings: clarify on some functions and make some of them miniscule Paper <mrpapersonic@gmail.com> parents: 98 diff changeset	126 while (string.find("<") != std::string::npos) {
503bc1547d49 strings: clarify on some functions and make some of them miniscule Paper <mrpapersonic@gmail.com> parents: 98 diff changeset	127 auto startpos = string.find("<");
503bc1547d49 strings: clarify on some functions and make some of them miniscule Paper <mrpapersonic@gmail.com> parents: 98 diff changeset	128 auto endpos = string.find(">") + 1;
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	129
99 503bc1547d49 strings: clarify on some functions and make some of them miniscule Paper <mrpapersonic@gmail.com> parents: 98 diff changeset	130 if (endpos != std::string::npos)
503bc1547d49 strings: clarify on some functions and make some of them miniscule Paper <mrpapersonic@gmail.com> parents: 98 diff changeset	131 string.erase(startpos, endpos - startpos);
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	132 }
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	133 }
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	134
98 582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	135 /* e.g. "<" for "<" */
260 dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	136 void ParseHtmlEntities(std::string& string) {
98 582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	137 const std::unordered_map<std::string, std::string> map = {
274 f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	138 {"<", "<"},
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	139 {"&rt;", ">"},
258 862d0d8619f6 : HUUUGE changes Paper <paper@paper.us.eu.org>* parents: 250 diff changeset	140 {" ", "\xA0"},
274 f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	141 {"&", "&"},
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	142 {""", "\""},
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	143 {"'", "'"},
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	144 {"¢", "¢"},
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	145 {"£", "£"},
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	146 {"€", "€"},
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	147 {"¥", "¥"},
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	148 {"©", "©"},
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	149 {"®", "®"},
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	150 {"×", "×"},
258 862d0d8619f6 : HUUUGE changes Paper <paper@paper.us.eu.org>* parents: 250 diff changeset	151 {"’", "’" } // Haibane Renmei, AniList
98 582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	152 };
582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	153
582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	154 for (const auto& item : map)
260 dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	155 ReplaceAll(string, item.first, item.second);
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	156 }
5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	157
99 503bc1547d49 strings: clarify on some functions and make some of them miniscule Paper <mrpapersonic@gmail.com> parents: 98 diff changeset	158 /* removes stupid HTML stuff */
260 dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	159 void TextifySynopsis(std::string& string) {
274 f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	160 #if 1
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	161 /* Just let Qt deal with it. */
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	162 QTextDocument text;
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	163 text.setHtml(Strings::ToQString(string));
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	164 string = Strings::ToUtf8String(text.toPlainText());
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	165 #else
f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	166 /* old implementation here... beware of jankiness */
260 dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	167 SanitizeLineEndings(string);
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	168 RemoveHtmlTags(string);
dd211ff68b36 pages/seasons: add initial functionality Paper <paper@paper.us.eu.org> parents: 258 diff changeset	169 ParseHtmlEntities(string);
274 f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	170 #endif
98 582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	171 }
582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	172
582b2fca1561 strings: parse HTML entities when reading synopsis, make the Paper <mrpapersonic@gmail.com> parents: 81 diff changeset	173 /* let Qt handle the heavy lifting of locale shit
221 53211cb1e7f5 library: add initial library stuff Paper <paper@paper.us.eu.org> parents: 211 diff changeset	174 * I don't want to deal with
258 862d0d8619f6 : HUUUGE changes Paper <paper@paper.us.eu.org>* parents: 250 diff changeset	175 */
15 cde8f67a7c7d : update, megacommit :) Paper <mrpapersonic@gmail.com>* parents: 9 diff changeset	176 std::string ToUpper(const std::string& string) {
187 9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	177 return ToUtf8String(session.config.locale.GetLocale().toUpper(ToQString(string)));
15 cde8f67a7c7d : update, megacommit :) Paper <mrpapersonic@gmail.com>* parents: 9 diff changeset	178 }
cde8f67a7c7d : update, megacommit :) Paper <mrpapersonic@gmail.com>* parents: 9 diff changeset	179
cde8f67a7c7d : update, megacommit :) Paper <mrpapersonic@gmail.com>* parents: 9 diff changeset	180 std::string ToLower(const std::string& string) {
187 9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	181 return ToUtf8String(session.config.locale.GetLocale().toLower(ToQString(string)));
15 cde8f67a7c7d : update, megacommit :) Paper <mrpapersonic@gmail.com>* parents: 9 diff changeset	182 }
cde8f67a7c7d : update, megacommit :) Paper <mrpapersonic@gmail.com>* parents: 9 diff changeset	183
62 4c6dd5999b39 : update Paper <mrpapersonic@gmail.com>* parents: 15 diff changeset	184 std::wstring ToWstring(const std::string& string) {
230 2f5a9247e501 torrents: implement download button Paper <paper@paper.us.eu.org> parents: 221 diff changeset	185 static std::wstring_convert<std::codecvt_utf8<wchar_t>> converter("", L"");
2f5a9247e501 torrents: implement download button Paper <paper@paper.us.eu.org> parents: 221 diff changeset	186
2f5a9247e501 torrents: implement download button Paper <paper@paper.us.eu.org> parents: 221 diff changeset	187 std::wstring wstr;
2f5a9247e501 torrents: implement download button Paper <paper@paper.us.eu.org> parents: 221 diff changeset	188 try {
2f5a9247e501 torrents: implement download button Paper <paper@paper.us.eu.org> parents: 221 diff changeset	189 wstr = converter.from_bytes(string);
2f5a9247e501 torrents: implement download button Paper <paper@paper.us.eu.org> parents: 221 diff changeset	190 } catch (std::range_error const& ex) {
2f5a9247e501 torrents: implement download button Paper <paper@paper.us.eu.org> parents: 221 diff changeset	191 std::cerr << "Failed to convert UTF-8 to wide string!" << std::endl;
2f5a9247e501 torrents: implement download button Paper <paper@paper.us.eu.org> parents: 221 diff changeset	192 }
2f5a9247e501 torrents: implement download button Paper <paper@paper.us.eu.org> parents: 221 diff changeset	193 return wstr;
62 4c6dd5999b39 : update Paper <mrpapersonic@gmail.com>* parents: 15 diff changeset	194 }
4c6dd5999b39 : update Paper <mrpapersonic@gmail.com>* parents: 15 diff changeset	195
64 fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	196 std::wstring ToWstring(const QString& string) {
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	197 std::wstring arr(string.size(), L'\0');
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	198 string.toWCharArray(&arr.front());
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	199 return arr;
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	200 }
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	201
62 4c6dd5999b39 : update Paper <mrpapersonic@gmail.com>* parents: 15 diff changeset	202 std::string ToUtf8String(const std::wstring& wstring) {
230 2f5a9247e501 torrents: implement download button Paper <paper@paper.us.eu.org> parents: 221 diff changeset	203 static std::wstring_convert<std::codecvt_utf8<wchar_t>> converter("", L"");
62 4c6dd5999b39 : update Paper <mrpapersonic@gmail.com>* parents: 15 diff changeset	204 return converter.to_bytes(wstring);
4c6dd5999b39 : update Paper <mrpapersonic@gmail.com>* parents: 15 diff changeset	205 }
4c6dd5999b39 : update Paper <mrpapersonic@gmail.com>* parents: 15 diff changeset	206
64 fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	207 std::string ToUtf8String(const QString& string) {
187 9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	208 const QByteArray ba = string.toUtf8();
77 6f7385bd334c : update Paper <mrpapersonic@gmail.com>* parents: 76 diff changeset	209 return std::string(ba.constData(), ba.size());
6f7385bd334c : update Paper <mrpapersonic@gmail.com>* parents: 76 diff changeset	210 }
6f7385bd334c : update Paper <mrpapersonic@gmail.com>* parents: 76 diff changeset	211
6f7385bd334c : update Paper <mrpapersonic@gmail.com>* parents: 76 diff changeset	212 std::string ToUtf8String(const QByteArray& ba) {
6f7385bd334c : update Paper <mrpapersonic@gmail.com>* parents: 76 diff changeset	213 return std::string(ba.constData(), ba.size());
64 fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	214 }
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	215
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	216 QString ToQString(const std::string& string) {
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	217 return QString::fromUtf8(string.c_str(), string.length());
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	218 }
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	219
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	220 QString ToQString(const std::wstring& wstring) {
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	221 return QString::fromWCharArray(wstring.c_str(), wstring.length());
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	222 }
fe719c109dbc : update Paper <mrpapersonic@gmail.com>* parents: 62 diff changeset	223
211 7cf53145de11 strings: use templates for ToInt, std::to_string -> Strings::ToUtf8String Paper <mrpapersonic@gmail.com> parents: 187 diff changeset	224 std::string ToUtf8String(const bool b) {
7cf53145de11 strings: use templates for ToInt, std::to_string -> Strings::ToUtf8String Paper <mrpapersonic@gmail.com> parents: 187 diff changeset	225 return b ? "true" : "false"; // lol
101 c537996cf67b : multitude of config changes Paper <mrpapersonic@gmail.com>* parents: 100 diff changeset	226 }
c537996cf67b : multitude of config changes Paper <mrpapersonic@gmail.com>* parents: 100 diff changeset	227
211 7cf53145de11 strings: use templates for ToInt, std::to_string -> Strings::ToUtf8String Paper <mrpapersonic@gmail.com> parents: 187 diff changeset	228 bool ToBool(const std::string& str, bool def) {
7cf53145de11 strings: use templates for ToInt, std::to_string -> Strings::ToUtf8String Paper <mrpapersonic@gmail.com> parents: 187 diff changeset	229 std::istringstream s(Strings::ToLower(str));
7cf53145de11 strings: use templates for ToInt, std::to_string -> Strings::ToUtf8String Paper <mrpapersonic@gmail.com> parents: 187 diff changeset	230 s >> std::boolalpha >> def;
116 254b1d2b7096 settings: add torrents page, make rss feed configurable Paper <mrpapersonic@gmail.com> parents: 114 diff changeset	231 return def;
254b1d2b7096 settings: add torrents page, make rss feed configurable Paper <mrpapersonic@gmail.com> parents: 114 diff changeset	232 }
254b1d2b7096 settings: add torrents page, make rss feed configurable Paper <mrpapersonic@gmail.com> parents: 114 diff changeset	233
211 7cf53145de11 strings: use templates for ToInt, std::to_string -> Strings::ToUtf8String Paper <mrpapersonic@gmail.com> parents: 187 diff changeset	234 /* util funcs */
114 ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	235 uint64_t HumanReadableSizeToBytes(const std::string& str) {
187 9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	236 static const std::unordered_map<std::string, uint64_t> bytes_map = {
273 f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	237 {"KB", 1000ull},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	238 {"MB", 1000000ull},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	239 {"GB", 1000000000ull},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	240 {"TB", 1000000000000ull},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	241 {"PB", 1000000000000000ull},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	242 {"KiB", 1ull << 10},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	243 {"MiB", 1ull << 20},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	244 {"GiB", 1ull << 30},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	245 {"TiB", 1ull << 40},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	246 {"PiB", 1ull << 50} /* surely we won't need more than this */
114 ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	247 };
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	248
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	249 for (const auto& suffix : bytes_map) {
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	250 if (str.find(suffix.first) != std::string::npos) {
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	251 try {
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	252 uint64_t size = std::stod(str) * suffix.second;
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	253 return size;
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	254 } catch (std::invalid_argument const& ex) {
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	255 continue;
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	256 }
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	257 }
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	258 }
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	259
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	260 return ToInt(str, 0);
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	261 }
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	262
273 f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	263 std::string BytesToHumanReadableSize(uint64_t bytes, int precision) {
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	264 #if QT_VERSION >= QT_VERSION_CHECK(5, 10, 0)
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	265 /* QLocale in Qt >= 5.10.0 has a function for this */
274 f6a756c19bfb anime_list.cc: use mutexes so we don't sex the stack Paper <paper@paper.us.eu.org> parents: 273 diff changeset	266 return Strings::ToUtf8String(session.config.locale.GetLocale().formattedDataSize(bytes, precision));
273 f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	267 #else
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	268 static const std::unordered_map<uint64_t, std::string> map = {
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	269 {1ull << 10, "KiB"},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	270 {1ull << 20, "MiB"},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	271 {1ull << 30, "GiB"},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	272 {1ull << 40, "TiB"},
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	273 {1ull << 50, "PiB"}
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	274 };
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	275
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	276 for (const auto& suffix : map) {
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	277 if (bytes / suffix.first < 1)
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	278 continue;
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	279
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	280 std::stringstream ss;
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	281 ss << std::setprecision(precision)
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	282 << (static_cast<double>(bytes) / suffix.first) << " "
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	283 << suffix.second;
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	284 return ss.str();
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	285 }
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	286
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	287 /* better luck next time */
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	288 return "0 bytes";
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	289 #endif
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	290 }
f31305b9f60a : various code safety changes Paper <paper@paper.us.eu.org>* parents: 264 diff changeset	291
264 9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	292 void RemoveLeadingChars(std::string& s, const char c) {
118 39521c47c7a3 : another huge megacommit, SORRY Paper <mrpapersonic@gmail.com>* parents: 116 diff changeset	293 s.erase(0, std::min(s.find_first_not_of(c), s.size() - 1));
114 ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	294 }
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	295
264 9a04802848c0 : improve multiple things Paper <paper@paper.us.eu.org>* parents: 260 diff changeset	296 void RemoveTrailingChars(std::string& s, const char c) {
114 ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	297 s.erase(s.find_last_not_of(c) + 1, std::string::npos);
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	298 }
ab191e28e69d : add initial torrent stuff Paper <mrpapersonic@gmail.com>* parents: 102 diff changeset	299
102 b315f3759c56 : big patch Paper <mrpapersonic@gmail.com>* parents: 101 diff changeset	300 bool BeginningMatchesSubstring(const std::string& str, const std::string& sub) {
b315f3759c56 : big patch Paper <mrpapersonic@gmail.com>* parents: 101 diff changeset	301 for (unsigned long long i = 0; i < str.length() && i < sub.length(); i++)
b315f3759c56 : big patch Paper <mrpapersonic@gmail.com>* parents: 101 diff changeset	302 if (str[i] != sub[i])
b315f3759c56 : big patch Paper <mrpapersonic@gmail.com>* parents: 101 diff changeset	303 return false;
187 9613d72b097e : multiple performance improvements Paper <mrpapersonic@gmail.com>* parents: 120 diff changeset	304
102 b315f3759c56 : big patch Paper <mrpapersonic@gmail.com>* parents: 101 diff changeset	305 return true;
b315f3759c56 : big patch Paper <mrpapersonic@gmail.com>* parents: 101 diff changeset	306 }
b315f3759c56 : big patch Paper <mrpapersonic@gmail.com>* parents: 101 diff changeset	307
9 5c0397762b53 INCOMPLETE: megacommit :) Paper <mrpapersonic@gmail.com> parents: diff changeset	308 } // namespace Strings

Mercurial > minori

annotate src/core/strings.cc @ 321:8141f409d52c