mirror/yacreader
1
0
Fork 0
mirror of https://github.com/YACReader/yacreader synced 2025-05-28 03:10:27 -04:00
Code Issues Actions Packages Projects Releases Wiki Activity

Improve search engine with new operators and new fields

Browse Source
This commit is contained in:
Luis Ángel San Martín 2023-05-21 20:29:13 +02:00
parent 117d65ffbe
commit 6d457be912
5 changed files with 150 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
CHANGELOG.md
View File

@ -10,6 +10,8 @@ Version counting is based on semantic versioning (Major.Feature.Patch)
* Add support for showing a "recently added/updated" indicator.
* Improved comic metadata dialog.
* Add textual tags support that can be queried through the search engine.
* Make = in the search engine work as : does.
* Add new operators to the search engine: exact match ==, <, >, <=, >=.
## 9.12

49
YACReaderLibrary/db/query_lexer.cpp
View File

@ -17,6 +17,14 @@ Token QueryLexer::next()
case '(':
case ')':
return single(Token::Type::opcode);
case ':':
return single(Token::Type::equal);
case '=':
return equal();
case '<':
return minor();
case '>':
return major();
case '"':
return quotedWord();
default:
@ -44,7 +52,7 @@ Token QueryLexer::word()
auto start = index;
get();
auto current = peek();
while (current != '\0' && !isSpace(current) && current != '"' && current != '(' && current != ')') {
while (current != '\0' && !isSpace(current) && current != '"' && current != '(' && current != ')' && current != ':' && current != '=' && current != '<' && current != '>') {
get();
current = peek();
}
@ -70,6 +78,45 @@ Token QueryLexer::quotedWord()
return Token(Token::Type::eof);
}
Token QueryLexer::minor()
{
auto start = index;
get();
auto current = peek();
if (current == '=') {
get();
return Token(Token::Type::minorOrEqual, input.substr(start, index - start));
}
return Token(Token::Type::minor, input.substr(start, index - start));
}
Token QueryLexer::major()
{
auto start = index;
get();
auto current = peek();
if (current == '=') {
get();
return Token(Token::Type::majorOrEqual, input.substr(start, index - start));
}
return Token(Token::Type::major, input.substr(start, index - start));
}
Token QueryLexer::equal()
{
auto start = index;
get();
auto current = peek();
if (current == '=') {
get();
return Token(Token::Type::exactEqual, input.substr(start, index - start));
}
return Token(Token::Type::equal, input.substr(start, index - start));
}
bool QueryLexer::isSpace(char c)
{
switch (c) {

9
YACReaderLibrary/db/query_lexer.h
View File

@ -11,6 +11,12 @@ public:
opcode,
word,
quotedWord,
equal, // =
exactEqual, // ==
minor,
major,
minorOrEqual,
majorOrEqual,
undefined
};
@ -50,6 +56,9 @@ private:
Token single(Token::Type type);
Token word();
Token quotedWord();
Token minor();
Token major();
Token equal();
bool isSpace(char c);
};

100
YACReaderLibrary/db/query_parser.cpp
View File

@ -8,19 +8,25 @@
#include <QsLog.h>
const std::map<QueryParser::FieldType, std::vector<std::string>> QueryParser::fieldNames {
{ FieldType::numeric, { "numpages", "count", "arccount", "alternateCount" } },
// TODO_METADATA support dates
{ FieldType::numeric, { "numpages", "count", "arccount", "alternateCount", "rating" } },
{ FieldType::text, { "number", "arcnumber", "title", "volume", "storyarc", "genere", "writer", "penciller", "inker", "colorist", "letterer", "coverartist", "publisher", "format", "agerating", "synopsis", "characters", "notes", "editor", "imprint", "teams", "locations", "series", "alternateSeries", "alternateNumber", "languageISO", "seriesGroup", "mainCharacterOrTeam", "review", "tags" } },
{ FieldType::boolean, { "isbis", "color", "read" } },
{ FieldType::boolean, { "color", "read", "edited", "hasBeenOpened" } },
{ FieldType::date, { "date", "added", "lastTimeOpened" } },
{ FieldType::filename, { "filename" } },
{ FieldType::folder, { "folder" } },
{ FieldType::booleanFolder, { "completed", "finished" } }, // TODO_METADTA include new folder fields, e.g. type
{ FieldType::enumField, { "type" } }
{ FieldType::enumField, { "type" } },
{ FieldType::enumFieldFolder, { "foldertype" } }
};
int QueryParser::TreeNode::buildSqlString(std::string &sqlString, int bindPosition) const
{
if (t == "token") {
// TODO: add some semantic checks, not all operators apply to all fields
// TODO: add support for == for an exact comparison
// TODO: try to add support for <,>,<=,>= for number, even if it's a string now maybe it can be done
if (t == "expression") {
++bindPosition;
if (toLower(children[0].t) == "all") {
sqlString += "(";
@ -29,7 +35,15 @@ int QueryParser::TreeNode::buildSqlString(std::string &sqlString, int bindPositi
}
sqlString += "UPPER(c.filename) LIKE UPPER(:bindPosition" + std::to_string(bindPosition) + ") OR ";
sqlString += "UPPER(f.name) LIKE UPPER(:bindPosition" + std::to_string(bindPosition) + ")) ";
} else if (isIn(fieldType(children[0].t), { FieldType::numeric, FieldType::boolean, FieldType::enumField })) {
} else if (isIn(fieldType(children[0].t), { FieldType::numeric })) {
std::string sqlOperator;
if (expOperator == ":" || expOperator == "=" || expOperator == "==") {
sqlOperator = "=";
} else {
sqlOperator = expOperator;
}
sqlString += "ci." + children[0].t + " " + sqlOperator + " :bindPosition" + std::to_string(bindPosition) + " ";
} else if (isIn(fieldType(children[0].t), { FieldType::boolean, FieldType::enumField })) {
sqlString += "ci." + children[0].t + " = :bindPosition" + std::to_string(bindPosition) + " ";
} else if (fieldType(children[0].t) == FieldType::filename) {
sqlString += "(UPPER(c." + children[0].t + ") LIKE UPPER(:bindPosition" + std::to_string(bindPosition) + ")) ";
@ -37,8 +51,23 @@ int QueryParser::TreeNode::buildSqlString(std::string &sqlString, int bindPositi
sqlString += "(UPPER(f.name) LIKE UPPER(:bindPosition" + std::to_string(bindPosition) + ")) ";
} else if (fieldType(children[0].t) == FieldType::booleanFolder) {
sqlString += "f." + children[0].t + " = :bindPosition" + std::to_string(bindPosition) + " ";
} else if (fieldType(children[0].t) == FieldType::enumFieldFolder) {
if (children[0].t == "foldertype") {
sqlString += "f.type = :bindPosition" + std::to_string(bindPosition) + " ";
} else {
sqlString += "f." + children[0].t + " = :bindPosition" + std::to_string(bindPosition) + " ";
}
} else {
sqlString += "(UPPER(ci." + children[0].t + ") LIKE UPPER(:bindPosition" + std::to_string(bindPosition) + ")) ";
if (expOperator == "=" || expOperator == ":" || expOperator == "") {
sqlString += "(UPPER(ci." + children[0].t + ") LIKE UPPER(:bindPosition" + std::to_string(bindPosition) + ")) ";
} else {
if (expOperator == "==") {
sqlString += "(UPPER(ci." + children[0].t + ") = UPPER(:bindPosition" + std::to_string(bindPosition) + ")) ";
} else {
// support for <,>,<=,>= in text fields makes sense for number, arcNumber, alternateNumber, but (TODO) the syntax won't prevent other fields from using this operators
sqlString += "(CAST(ci." + children[0].t + " as REAL) " + expOperator + " CAST(:bindPosition" + std::to_string(bindPosition) + " as REAL)) ";
}
}
}
} else if (t == "not") {
sqlString += "(NOT ";
@ -57,7 +86,7 @@ int QueryParser::TreeNode::buildSqlString(std::string &sqlString, int bindPositi
int QueryParser::TreeNode::bindValues(QSqlQuery &selectQuery, int bindPosition) const
{
if (t == "token") {
if (t == "expression") {
std::string bind_string(":bindPosition" + std::to_string(++bindPosition));
if (isIn(fieldType(children[0].t), { FieldType::numeric })) {
selectQuery.bindValue(QString::fromStdString(bind_string), std::stoi(children[1].t));
@ -70,10 +99,10 @@ int QueryParser::TreeNode::bindValues(QSqlQuery &selectQuery, int bindPosition)
} else {
selectQuery.bindValue(QString::fromStdString(bind_string), std::stoi(value));
}
} else if ((isIn(fieldType(children[0].t), { FieldType::enumField }))) {
} else if ((isIn(fieldType(children[0].t), { FieldType::enumField, FieldType::enumFieldFolder }))) {
auto enumType = children[0].t;
auto value = toLower(children[1].t);
if (enumType == "type") {
if (enumType == "type" || enumType == "foldertype") {
if (value == "comic") {
selectQuery.bindValue(QString::fromStdString(bind_string), 0);
} else if (value == "manga") {
@ -89,7 +118,11 @@ int QueryParser::TreeNode::bindValues(QSqlQuery &selectQuery, int bindPosition)
selectQuery.bindValue(QString::fromStdString(bind_string), std::stoi(children[1].t));
}
} else {
selectQuery.bindValue(QString::fromStdString(bind_string), QString::fromStdString("%%" + children[1].t + "%%"));
if (expOperator == "=" || expOperator == ":" || expOperator == "") {
selectQuery.bindValue(QString::fromStdString(bind_string), QString::fromStdString("%%" + children[1].t + "%%"));
} else {
selectQuery.bindValue(QString::fromStdString(bind_string), QString::fromStdString(children[1].t));
}
}
} else if (t == "not") {
bindPosition = children[0].bindValues(selectQuery, bindPosition);
@ -172,6 +205,16 @@ void QueryParser::advance()
currentToken = lexer.next();
}
bool QueryParser::isOperatorToken(Token::Type type)
{
return type == Token::Type::equal ||
type == Token::Type::exactEqual ||
type == Token::Type::minor ||
type == Token::Type::minorOrEqual ||
type == Token::Type::major ||
type == Token::Type::majorOrEqual;
}
QueryParser::FieldType QueryParser::fieldType(const std::string &str)
{
for (const auto &names : fieldNames) {
@ -244,26 +287,39 @@ QueryParser::TreeNode QueryParser::locationExpression()
if (!isIn(tokenType(), { Token::Type::word, Token::Type::quotedWord })) {
throw std::invalid_argument("Invalid syntax. Expected a lookup name or a word");
}
return expression();
}
QueryParser::TreeNode QueryParser::expression()
{
if (tokenType() == Token::Type::word) {
auto left = token(true);
if (isOperatorToken(tokenType())) {
auto expOperator = token(true);
if (tokenType() != Token::Type::word && tokenType() != Token::Type::quotedWord) {
throw std::invalid_argument("missing right operand");
}
auto right = token(true);
return TreeNode("expression", { TreeNode(toLower(left), {}), TreeNode(right, {}) }, expOperator);
} else {
return TreeNode("expression", { TreeNode("all", {}), TreeNode(left, {}) });
}
}
return baseToken();
}
QueryParser::TreeNode QueryParser::baseToken()
{
if (tokenType() == Token::Type::quotedWord) {
return TreeNode("token", { TreeNode("all", {}), TreeNode(token(true), {}) });
return TreeNode("expression", { TreeNode("all", {}), TreeNode(token(true), {}) });
}
// TODO ":" should come from the lexer as a token
auto words(split(token(true), ':'));
if (words.size() > 1 && fieldType(words[0].toStdString()) != FieldType::unknown) {
auto loc(toLower(words[0].toStdString()));
words.erase(words.begin());
if (words.size() == 1 && tokenType() == Token::Type::quotedWord) {
return TreeNode("token", { TreeNode(loc, {}), TreeNode(token(true), {}) });
}
return TreeNode("token", { TreeNode(loc, {}), TreeNode(join(words, ":"), {}) });
if (tokenType() == Token::Type::word) {
return TreeNode("expression", { TreeNode("all", {}), TreeNode(token(true), {}) });
}
return TreeNode("token", { TreeNode("all", {}), TreeNode(join(words, ":"), {}) });
return TreeNode("expression", { TreeNode("all", {}), TreeNode(token(true), {}) });
}

19
YACReaderLibrary/db/query_parser.h
View File

@ -9,8 +9,8 @@
#include <vector>
#include <list>
#define SEARCH_FOLDERS_QUERY "SELECT DISTINCT f.id, f.parentId, f.name, f.path, f.finished, f.completed, f.numChildren, f.firstChildHash FROM folder f LEFT JOIN comic c ON (f.id = c.parentId) INNER JOIN comic_info ci ON (c.comicInfoId = ci.id) WHERE "
#define SEARCH_COMICS_QUERY "SELECT ci.number,ci.title,c.fileName,ci.numPages,c.id,c.parentId,c.path,ci.hash,ci.read,ci.isBis,ci.currentPage,ci.rating,ci.hasBeenOpened,ci.coverSizeRatio,ci.lastTimeOpened,ci.manga FROM comic c INNER JOIN comic_info ci ON (c.comicInfoId = ci.id) LEFT JOIN folder f ON (f.id == c.parentId) WHERE "
#define SEARCH_FOLDERS_QUERY "SELECT DISTINCT * FROM folder f LEFT JOIN comic c ON (f.id = c.parentId) INNER JOIN comic_info ci ON (c.comicInfoId = ci.id) WHERE "
#define SEARCH_COMICS_QUERY "SELECT ci.number,ci.title,c.fileName,ci.numPages,c.id,c.parentId,c.path,ci.hash,ci.read,ci.isBis,ci.currentPage,ci.rating,ci.hasBeenOpened,ci.date,ci.added,ci.type FROM comic c INNER JOIN comic_info ci ON (c.comicInfoId = ci.id) LEFT JOIN folder f ON (f.id == c.parentId) WHERE "
/**
* This class is used to generate an SQL query string from a search expression,
@ -25,7 +25,9 @@
* and_expression ::= not_expression [ [ 'and' ] and_expression ]
* not_expression ::= [ 'not' ] location_expression
* location_expression ::= base_token | ( '(' or_expression ')' )
* base_token ::= a sequence of letters and colons, perhaps quoted
* expression :: base_token | base_token 'operator' base_token
* operator :: [':' '=' '<' '>' '<=' '=>']
* base_token ::= a sequence of letters, perhaps quoted
*
* Usage Example:
* QSqlQuery selectQuery(db);
@ -47,9 +49,10 @@ public:
struct TreeNode {
std::string t;
std::vector<TreeNode> children;
std::string expOperator;
explicit TreeNode(std::string t, std::vector<TreeNode> children)
: t(t), children(children)
explicit TreeNode(std::string t, std::vector<TreeNode> children, std::string expOperator = "")
: t(t), children(children), expOperator(expOperator)
{
}
@ -78,6 +81,8 @@ private:
return std::find(v.begin(), v.end(), e) != v.end();
}
bool isOperatorToken(Token::Type type);
enum class FieldType { unknown,
numeric,
text,
@ -86,7 +91,8 @@ private:
folder,
booleanFolder,
filename,
enumField };
enumField,
enumFieldFolder };
static FieldType fieldType(const std::string &str);
static std::string join(const QStringList &strings, const std::string &delim);
@ -96,6 +102,7 @@ private:
TreeNode andExpression();
TreeNode notExpression();
TreeNode locationExpression();
TreeNode expression();
TreeNode baseToken();
static const std::map<FieldType, std::vector<std::string>> fieldNames;