Cockatrice/oracle/src/oracleimporter.cpp
BruebachL c011ea7ceb
[Oracle] Parse sets lazily to slash importer peak memory (#7217)
* [Oracle] Parse sets lazily to slash importer peak memory

- Add a raw JSON scanner that splits the document into per-set byte ranges
  without materializing the JSON tree
- Keep only the raw document bytes and parse one set at a time in startImport
- Take readSetsFromByteArray by value so the wizard's buffer is moved, not copied
- Clear the retained raw data in releaseSetData()/clear()
- Cover the scanner and lazy parsing with tests

Took 2 minutes

* [Oracle] Fix nesting-depth cap, tolerate unescaped control chars, lazy-parse review fixes

---------

Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>
2026-09-05 20:35:28 +02:00

645 lines
26 KiB
C++
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

#include "oracleimporter.h"
#include "libcockatrice/interfaces/noop_card_preference_provider.h"
#include "libcockatrice/interfaces/noop_card_set_priority_controller.h"
#include "parsehelpers.h"
#include <QDebug>
#include <QJsonDocument>
#include <QJsonObject>
#include <QJsonParseError>
#include <QRegularExpression>
#include <QSet>
#include <algorithm>
#include <climits>
#include <libcockatrice/card/database/parser/cockatrice_xml_4.h>
#include <libcockatrice/card/relation/card_relation.h>
static const QList<AllowedCount> kConstructedCounts = {{4, "legal"}, {0, "banned"}};
static const QList<AllowedCount> kSingletonCounts = {{1, "legal"}, {0, "banned"}};
SplitCardPart::SplitCardPart(const QString &_name,
const QString &_text,
const QHash<QString, QString> &_properties,
const PrintingInfo &_printingInfo)
: name(_name), text(_text), properties(_properties), printingInfo(_printingInfo)
{
}
const QRegularExpression OracleImporter::formatRegex = QRegularExpression("^format-");
OracleImporter::OracleImporter(QObject *parent) : QObject(parent)
{
}
static CardSet::Priority getSetPriority(const QString &setType, const QString &shortName)
{
if (!setTypePriorities.contains(setType.toLower())) {
qDebug() << "warning: Set type" << setType << "unrecognized for prioritization";
}
CardSet::Priority priority = setTypePriorities.value(setType.toLower(), CardSet::PriorityOther);
if (nonEnglishSets.contains(shortName)) {
priority = CardSet::PriorityLowest;
}
return priority;
}
bool OracleImporter::readSetsFromByteArray(QByteArray data)
{
RawJson::ScanError error;
const QList<RawJson::SetRange> ranges = RawJson::scanSetRanges(data, &error);
if (error.isError()) {
qDebug() << "error: RawJson::scanSetRanges():" << error.message;
return false;
}
QList<SetToDownload> newSetList;
newSetList.reserve(ranges.size());
for (const RawJson::SetRange &range : ranges) {
QString shortName = range.code.toUpper();
QString longName = range.name;
QString setType = range.type;
QDate releaseDate = QDate::fromString(range.releaseDate, Qt::ISODate);
CardSet::Priority priority = getSetPriority(setType, shortName);
// capitalize set type
if (setType.length() > 0) {
// basic grammar for words that aren't capitalized, like in "From the Vault"
static const QStringList noCapitalize = {"the", "a", "an", "on", "to", "for",
"of", "in", "and", "with", "or"};
QStringList words = setType.split("_");
setType.clear();
bool first = false;
for (auto &item : words) {
if (first && noCapitalize.contains(item)) {
setType += item + QString(" ");
} else {
setType += item[0].toUpper() + item.mid(1) + QString(" ");
first = true;
}
}
setType = setType.trimmed();
}
SetToDownload set(shortName, longName, priority, setType, releaseDate);
set.setRawRange(range.dataRange);
newSetList.append(set);
}
std::sort(newSetList.begin(), newSetList.end());
if (newSetList.isEmpty()) {
return false;
}
allSets = newSetList;
rawSetsData = std::move(data);
return true;
}
static QString getMainCardType(const QStringList &typeList)
{
if (typeList.isEmpty()) {
return {};
}
static const QStringList typePriority = {"Planeswalker", "Creature", "Land", "Sorcery",
"Instant", "Artifact", "Enchantment"};
for (const auto &type : typePriority) {
if (typeList.contains(type)) {
return type;
}
}
return typeList.first();
}
/**
* Sorts and deduplicates the color chars in the string by WUBRG order.
*
* @param colors The string containing the color chars. Will be modified in-place
*/
static void sortAndReduceColors(QString &colors)
{
// sort
static const QHash<QChar, unsigned int> colorOrder{{'W', 0}, {'U', 1}, {'B', 2}, {'R', 3}, {'G', 4}};
std::sort(colors.begin(), colors.end(),
[](const QChar a, const QChar b) { return colorOrder.value(a, INT_MAX) < colorOrder.value(b, INT_MAX); });
// reduce
auto last = std::unique(colors.begin(), colors.end());
colors.erase(last, colors.end());
}
CardInfoPtr OracleImporter::addCard(QString name,
const QString &text,
bool isToken,
QHash<QString, QString> properties,
const QList<CardRelation *> &relatedCards,
const PrintingInfo &printingInfo)
{
// Workaround for card name weirdness
name = name.replace("Æ", "AE");
name = name.replace("", "'");
auto existingIt = cards.constFind(name);
if (existingIt != cards.constEnd()) {
CardInfoPtr card = existingIt.value();
card->addToSet(printingInfo.getSet(), printingInfo);
// Only merge legalities when the card has none yet, so multi-format
// printings don't overwrite each other's legality lists.
if (card->getProperties().filter(formatRegex).empty()) {
card->combineLegalities(properties);
}
return card;
}
// Remove {} around mana costs, except if it's split cost
QString manacost = properties.value("manacost");
if (!manacost.isEmpty()) {
QStringList symbols = manacost.split("}");
QString formattedCardCost;
for (QString symbol : symbols) {
static const auto manaCostPattern = QRegularExpression("[0-9WUBGRP]/[0-9WUBGRP]");
if (symbol.contains(manaCostPattern)) {
symbol.append("}");
} else {
symbol.remove(QChar('{'));
}
formattedCardCost.append(symbol);
}
properties.insert("manacost", formattedCardCost);
}
// fix colors
QString allColors = properties.value("colors");
if (allColors.size() > 1) {
sortAndReduceColors(allColors);
properties.insert("colors", allColors);
}
QString allColorIdent = properties.value("coloridentity");
if (allColorIdent.size() > 1) {
sortAndReduceColors(allColorIdent);
properties.insert("coloridentity", allColorIdent);
}
// DETECT CARD POSITIONING INFO
QString layoutVal = properties.value("layout");
bool landscapeOrientation =
properties.value("maintype") == "Battle" || layoutVal == "split" || layoutVal == "planar";
// cards that enter the field tapped
bool cipt = parseCipt(name, text) || landscapeOrientation;
// table row
int tableRow = 1;
QString mainCardType = properties.value("maintype");
if (mainCardType == "Land") {
tableRow = 0;
} else if (mainCardType == "Sorcery" || mainCardType == "Instant") {
tableRow = 3;
} else if (mainCardType == "Creature") {
tableRow = 2;
}
// card side
QString side = properties.value("side") == "b" ? "back" : "front";
properties.insert("side", side);
// upsideDown (flip cards)
QString layout = properties.value("layout");
bool upsideDown = layout == "flip" && side == "back";
// insert the card and its properties
SetToPrintingsMap setsInfo;
setsInfo[printingInfo.getSet()->getShortName()].append(printingInfo);
CardInfo::UiAttributes attributes = {cipt, landscapeOrientation, tableRow, upsideDown};
CardInfoPtr newCard =
CardInfo::newInstance(name, text, isToken, properties, relatedCards, {}, setsInfo, attributes);
if (name.isEmpty()) {
qDebug() << "warning: an empty card was added to set" << printingInfo.getSet()->getShortName();
}
cards.insert(name, newCard);
return newCard;
}
static QString getJsonString(const QJsonObject &obj, const QString &key)
{
// QVariant coerces numbers and booleans to text, while QJsonValue::toString()
// returns a null string for them — some MTGJSON fields (manaValue,
// convertedManaCost, isOnlineOnly, isRebalanced) carry those types.
return obj.value(key).toVariant().toString();
}
int OracleImporter::importCardsFromSet(const CardSetPtr &currentSet, const QJsonArray &cardsList)
{
// mtgjson name => xml name
static const QMap<QString, QString> cardProperties{
{"manaCost", "manacost"}, {"manaValue", "cmc"}, {"type", "type"},
{"loyalty", "loyalty"}, {"layout", "layout"}, {"side", "side"},
{"convertedManaCost", "cmc"}, // old name for manaValue, for backwards compatibility
};
// mtgjson name => xml name
static const QMap<QString, QString> setInfoProperties{{"number", "num"},
{"rarity", "rarity"},
{"isOnlineOnly", "isOnlineOnly"},
{"isRebalanced", "isRebalanced"},
{"artist", "artist"}};
// mtgjson name => xml name
static const QMap<QString, QString> identifierProperties{{"multiverseId", "muid"}, {"scryfallId", "uuid"}};
static const QString ptSeparator = "/";
static constexpr bool isToken = false;
static const QSet<QString> setsWithCardsWithSameNameButDifferentText = {"UST"};
int numCards = 0;
// Keeps track of any split card faces encountered so far
QMap<QString, QPair<QList<SplitCardPart>, QString>> splitCards;
// Keeps track of all names encountered so far
QSet<QString> allNameProps;
for (const QJsonValue &cardVal : cardsList) {
QJsonObject card = cardVal.toObject();
/* Currently used layouts are:
* augment, double_faced_token, flip, host, leveler, meld, normal, planar,
* saga, scheme, split, token, transform, vanguard
*/
QString layout = getJsonString(card, "layout");
// don't import tokens from the json file
if (layout == "token") {
continue;
}
// normal cards handling
QString name = getJsonString(card, "name");
QString text = getJsonString(card, "text");
QString faceName = getJsonString(card, "faceName");
if (faceName.isEmpty()) {
faceName = name;
}
// card properties
QHash<QString, QString> properties;
for (auto i = cardProperties.cbegin(), end = cardProperties.cend(); i != end; ++i) {
QString propertyValue = getJsonString(card, i.key());
if (!propertyValue.isEmpty()) {
properties.insert(i.value(), propertyValue);
}
}
// per-set properties
QHash<QString, QString> printingProps;
for (auto i = setInfoProperties.cbegin(), end = setInfoProperties.cend(); i != end; ++i) {
QString propertyValue = getJsonString(card, i.key());
if (!propertyValue.isEmpty()) {
printingProps.insert(i.value(), propertyValue);
}
}
// handle flavorNames specially due to double-faced cards
QString faceFlavorName = getJsonString(card, "faceFlavorName");
QString flavorName = !faceFlavorName.isEmpty() ? faceFlavorName : getJsonString(card, "flavorName");
if (!flavorName.isEmpty()) {
printingProps.insert("flavorName", flavorName);
}
// Identifiers
QJsonObject identifiers = card.value("identifiers").toObject();
for (auto i = identifierProperties.cbegin(), end = identifierProperties.cend(); i != end; ++i) {
QString propertyValue = getJsonString(identifiers, i.key());
if (!propertyValue.isEmpty()) {
printingProps.insert(i.value(), propertyValue);
}
}
PrintingInfo printingInfo(currentSet, LazyPropertiesHash(printingProps));
QString numComponent;
const QString numProperty = printingInfo.getProperty("num");
const QChar lastChar = numProperty.isEmpty() ? QChar() : numProperty.back();
// Un-Sets do some wonky stuff. Split up these cards as individual entries.
// these cards will have a num with a letter (abc) behind it, put that letter into the name
if (setsWithCardsWithSameNameButDifferentText.contains(currentSet->getShortName()) &&
allNameProps.contains(faceName) && layout == "normal" && lastChar.isLetter()) {
numComponent = " (" + QString(lastChar).toLower() + ")";
}
allNameProps.insert(faceName);
// special handling properties
QString colors;
for (const QJsonValue &color : card.value("colors").toArray()) {
colors += color.toString();
}
if (!colors.isEmpty()) {
properties.insert("colors", colors);
}
QString colorIdentity;
for (const QJsonValue &color : card.value("colorIdentity").toArray()) {
colorIdentity += color.toString();
}
if (!colorIdentity.isEmpty()) {
properties.insert("coloridentity", colorIdentity);
}
const auto &mainCardType = getMainCardType(card.value("types").toVariant().toStringList());
if (mainCardType.isEmpty()) {
qDebug() << "warning: no mainCardType for card:" << name;
} else {
properties.insert("maintype", mainCardType);
}
// Depending on whether power and/or toughness are present, the format
// is either P/T (most common), P (no toughness), or /T (no power).
QString power = getJsonString(card, "power");
QString toughness = getJsonString(card, "toughness");
if (toughness.isEmpty() && !power.isEmpty()) {
properties.insert("pt", power);
} else if (!toughness.isEmpty()) {
properties.insert("pt", power + ptSeparator + toughness);
}
auto legalities = card.value("legalities").toObject();
for (auto i = legalities.constBegin(), end = legalities.constEnd(); i != end; ++i) {
properties.insert(QString("format-%1").arg(i.key()), i.value().toString().toLower());
}
// split cards are considered a single card, enqueue for later merging
if (layout == "split" || layout == "aftermath" || layout == "adventure" || layout == "prepare") {
auto _faceName = getJsonString(card, "faceName");
SplitCardPart split(_faceName, text, properties, printingInfo);
auto found_iter = splitCards.find(name + numProperty);
if (found_iter == splitCards.end()) {
splitCards.insert(name + numProperty, {{split}, name});
} else {
found_iter->first.append(split);
}
} else {
// relations
QList<CardRelation *> relatedCards;
// add other face for split cards as card relation
if (!getJsonString(card, "side").isEmpty()) {
auto faceManaValue = getJsonString(card, "faceManaValue");
if (faceManaValue.isEmpty()) {
// check the old name for the property, for backwards compatibility purposes
faceManaValue = getJsonString(card, "faceConvertedManaCost");
}
properties["cmc"] = faceManaValue;
if (layout == "meld") { // meld cards don't work
static const QRegularExpression meldNameRegex{"then meld them into ([^\\.]*)"};
QString additionalName = meldNameRegex.match(text).captured(1);
if (!additionalName.isNull()) {
relatedCards.append(new CardRelation(additionalName, CardRelationType::TransformInto));
}
} else {
for (const QString &additionalName : name.split(" // ")) {
if (additionalName != faceName) {
relatedCards.append(new CardRelation(additionalName, CardRelationType::TransformInto));
}
}
}
name = faceName;
}
// mtgjson related cards
QJsonObject givenRelated = card.value("relatedCards").toObject();
if (!givenRelated.isEmpty()) {
// conjured cards from a spellbook
QJsonArray spellbook = givenRelated.value("spellbook").toArray();
if (!spellbook.isEmpty()) {
for (const QJsonValue &spbkVal : spellbook) {
relatedCards.append(new CardRelation(spbkVal.toString(), CardRelationType::DoesNotAttach, false,
false, 1, true));
}
}
}
CardInfoPtr newCard =
addCard(name + numComponent, text, isToken, std::move(properties), relatedCards, printingInfo);
numCards++;
}
}
// split cards handling
static const QString splitCardPropSeparator = QString(" // ");
static const QString splitCardTextSeparator = QString("\n\n---\n\n");
QList<QPair<QList<SplitCardPart>, QString>> partsAndNames = splitCards.values();
for (auto [splitCardParts, name] : partsAndNames) {
QString text;
QHash<QString, QString> properties;
PrintingInfo printingInfo;
for (const SplitCardPart &tmp : splitCardParts) {
if (!text.isEmpty()) {
text.append(splitCardTextSeparator);
}
text.append(tmp.getText());
if (properties.isEmpty()) {
properties = tmp.getProperties();
printingInfo = tmp.getPrintingInfo();
} else {
const QHash<QString, QString> &tmpProps = tmp.getProperties();
for (auto i = tmpProps.cbegin(), end = tmpProps.cend(); i != end; ++i) {
QString prop = i.key();
QString originalPropertyValue = properties.value(prop);
QString thisCardPropertyValue = i.value();
if (!thisCardPropertyValue.isEmpty() && originalPropertyValue != thisCardPropertyValue) {
if (originalPropertyValue.isEmpty()) { // don't create //es if one field is empty
properties.insert(prop, thisCardPropertyValue);
} else if (prop == "colors" || prop == "coloridentity") { // the card is both colors
properties.insert(prop, originalPropertyValue + thisCardPropertyValue);
} else if (prop == "maintype") { // don't create maintypes with //es in them
continue;
} else {
properties.insert(prop,
originalPropertyValue + splitCardPropSeparator + thisCardPropertyValue);
}
}
}
}
}
CardInfoPtr newCard = addCard(name, text, isToken, std::move(properties), {}, printingInfo);
numCards++;
}
return numCards;
}
static FormatRulesNameMap buildDefaultMagicFormats()
{
// Predefined common exceptions
CardCondition superTypeIsBasic;
superTypeIsBasic.field = "type";
superTypeIsBasic.matchType = "regex";
superTypeIsBasic.value = R"(\bBasic\b[^—]+\bLand\b)";
ExceptionRule basicLands;
basicLands.conditions.append(superTypeIsBasic);
CardCondition anyNumberAllowed;
anyNumberAllowed.field = "text";
anyNumberAllowed.matchType = "contains";
anyNumberAllowed.value = "A deck can have any number of";
ExceptionRule mayContainAnyNumber;
mayContainAnyNumber.conditions.append(anyNumberAllowed);
FormatRulesNameMap defaultFormatRulesNameMap;
// ----------------- Helper lambda to create format -----------------
auto makeFormat = [&](const QString &name, int minDeck = 60, int maxDeck = -1, int maxSideboardSize = 15,
const QList<AllowedCount> &allowedCounts = kConstructedCounts) -> FormatRulesPtr {
FormatRulesPtr f(new FormatRules);
f->formatName = name;
f->allowedCounts = allowedCounts;
f->minDeckSize = minDeck;
f->maxDeckSize = maxDeck;
f->maxSideboardSize = maxSideboardSize;
f->exceptions.append(basicLands);
f->exceptions.append(mayContainAnyNumber);
defaultFormatRulesNameMap.insert(name.toLower(), f);
return f;
};
// ----------------- Standard formats -----------------
makeFormat("Standard");
makeFormat("Modern");
makeFormat("Legacy");
makeFormat("Pioneer");
makeFormat("Historic");
makeFormat("Timeless");
makeFormat("Future");
makeFormat("OldSchool");
makeFormat("Premodern");
makeFormat("Pauper");
makeFormat("Penny");
// ----------------- Singleton formats -----------------
makeFormat("Commander", 100, 100, 15, kSingletonCounts);
makeFormat("Duel", 100, 100, 15, kSingletonCounts);
makeFormat("Brawl", 60, 60, 15, kSingletonCounts);
makeFormat("StandardBrawl", 60, 60, 15, kSingletonCounts);
makeFormat("Oathbreaker", 60, 60, 15, kSingletonCounts);
makeFormat("PauperCommander", 100, 100, 15, kSingletonCounts);
makeFormat("Predh", 100, 100, 15, kSingletonCounts);
// ----------------- Restricted formats -----------------
makeFormat("Vintage", 60, -1, 15, {{4, "legal"}, {1, "restricted"}, {0, "banned"}});
return defaultFormatRulesNameMap;
}
const FormatRulesNameMap &OracleImporter::createDefaultMagicFormats()
{
static const FormatRulesNameMap cached = buildDefaultMagicFormats();
return cached;
}
int OracleImporter::startImport()
{
static ICardSetPriorityController *noOpController = new NoopCardSetPriorityController();
// Pre-allocate the cards hash to avoid rehashing during import. Keys are
// distinct card names while raw ranges only count printings (AllPrintings
// ~100k printings vs ~35k names), so this over-reserves somewhat; an exact
// distinct-name count would require eagerly parsing, which the lazy reader
// deliberately avoids. It's a capacity hint, so the overshoot is harmless.
int estimatedCards = 0;
for (const SetToDownload &curSetToParse : allSets) {
estimatedCards += curSetToParse.getRawRange().cardCount;
}
cards.reserve(estimatedCards);
// add an empty set for tokens
CardSetPtr tokenSet =
CardSet::newInstance(noOpController, CardSet::TOKENS_SETNAME, tr("Dummy set containing tokens"), "Tokens");
sets.insert(CardSet::TOKENS_SETNAME, tokenSet);
int setIndex = 0;
for (const SetToDownload &curSetToParse : allSets) {
CardSetPtr newSet = CardSet::newInstance(noOpController, curSetToParse.getShortName(),
curSetToParse.getLongName(), curSetToParse.getSetType(),
curSetToParse.getReleaseDate(), curSetToParse.getPriority());
// parse only this set's slice of the raw document so the whole JSON tree is
// never kept in memory at once
const RawJson::SetDataRange &rawRange = curSetToParse.getRawRange();
const qsizetype rangeEnd = rawRange.start + rawRange.length;
if (rawRange.start < 0 || rawRange.length <= 0 || rangeEnd > rawSetsData.size()) {
// rawSetsData is cleared by releaseSetData() while SetToDownload copies
// taken from getSets() keep their ranges, and nothing else enforces the
// pairing — so never index past the buffer on stale/mismatched ranges.
qWarning() << "error: out-of-bounds raw range for set" << curSetToParse.getShortName() << "skipping";
++setIndex;
emit setIndexChanged(0, setIndex, curSetToParse.getLongName());
continue;
}
// sliced() shares the buffer instead of deep-copying the slice; the largest
// sets in AllPrintings are tens of MB, so the copy is worth avoiding here.
const QByteArray setBytes = rawSetsData.sliced(rawRange.start, rawRange.length);
QJsonParseError parseError;
const QJsonDocument setDoc = QJsonDocument::fromJson(setBytes, &parseError);
if (parseError.error != QJsonParseError::NoError) {
qWarning() << "error: parsing card data for set" << curSetToParse.getShortName() << ":"
<< parseError.errorString();
++setIndex;
// Keep the progress accounting honest: a set that failed to parse
// still advanced the index, so report it (with zero imported cards)
// rather than letting SaveSetsPage's bar stall per failed set.
emit setIndexChanged(0, setIndex, curSetToParse.getLongName());
continue;
}
// Only add the set to the database once its slice parsed cleanly;
// a set that fails here must not persist as an empty set in cards.xml.
if (!sets.contains(newSet->getShortName())) {
sets.insert(newSet->getShortName(), newSet);
}
const QJsonArray setCards = setDoc.object().value("cards").toArray();
int numCardsInSet = importCardsFromSet(newSet, setCards);
++setIndex;
emit setIndexChanged(numCardsInSet, setIndex, curSetToParse.getLongName());
}
emit setIndexChanged(0, setIndex, QString());
// total number of sets
return setIndex;
}
bool OracleImporter::saveToFile(const QString &fileName, const QString &sourceUrl, const QString &sourceVersion)
{
CockatriceXml4Parser parser(new NoopCardPreferenceProvider(), new NoopCardSetPriorityController());
return parser.saveToFile(createDefaultMagicFormats(), sets, cards, fileName, sourceUrl, sourceVersion);
}
void OracleImporter::releaseSetData()
{
allSets.clear();
rawSetsData.clear();
}
void OracleImporter::clear()
{
sets.clear();
cards.clear();
allSets.clear();
rawSetsData.clear();
}