XMB Improvements, parse JS into XMB, make strings more efficient.

XMB format is bumped to 4, invalidating all cached files. The
differences are:
 - element/attribute names are stored after the elements themselves, and
not before. This allows writing XMB data in one pass instead of two.
 - names themselves becomes offsets (instead of arbitrary integers),
making getting the string from the int name much more efficient.

XMBFile is renamed to XMBData to clarify that it does not, in fact,
refer to a file on disk.

XMBData::GetElementString is also changed to return a const char*, thus
not creating an std::string. A string_view version is added where
convenient.

The XML->XMB and JS->XMB conversion functions and the corresponding
storage are moved to `ps/XMB`, since that format doesn't particularly
relate to XML. CXeromyces becomes lighter and more focused as a result.
The XML->XMB conversion also benefits from the above streamlining.

Note that in a few cases, string_view gets printed to CLogger via
data(), which is generally not legal, but we know that the strings are
null-terminated here. Our libfmt (version 4) doesn't support
string_view, that would be v5.

Differential Revision: https://code.wildfiregames.com/D3909
This was SVN commit r25375.
This commit is contained in:
wraitii
2021-05-04 13:02:34 +00:00
parent a1010b83d3
commit cdd75deafb
20 changed files with 764 additions and 415 deletions
+14 -205
View File
@@ -28,6 +28,7 @@
#include "ps/CacheLoader.h"
#include "ps/CLogger.h"
#include "ps/Filesystem.h"
#include "RelaxNG.h"
#include "Xeromyces.h"
@@ -120,13 +121,17 @@ PSRETURN CXeromyces::Load(const PIVFS& vfs, const VfsPath& filename, const std::
validatorGrammarHash = GetValidator(validatorName).GetGrammarHash();
}
VfsPath xmbPath;
Status ret = cacheLoader.TryLoadingCached(filename, validatorGrammarHash, XMBVersion, xmbPath);
Status ret = cacheLoader.TryLoadingCached(filename, validatorGrammarHash, XMBStorage::XMBVersion, xmbPath);
if (ret == INFO::OK)
{
// Found a cached XMB - load it
if (ReadXMBFile(vfs, xmbPath))
if (m_Data.ReadFromFile(vfs, xmbPath))
{
if(!Initialise(m_Data))
return PSRETURN_Xeromyces_XMLParseError;
return PSRETURN_OK;
}
// If this fails then we'll continue and (re)create the loose cache -
// this failure legitimately happens due to partially-written XMB files.
}
@@ -184,43 +189,20 @@ PSRETURN CXeromyces::ConvertFile(const PIVFS& vfs, const VfsPath& filename, cons
}
}
WriteBuffer writeBuffer;
CreateXMB(doc, writeBuffer);
m_Data.LoadXMLDoc(doc);
xmlFreeDoc(doc);
// Save the file to disk, so it can be loaded quickly next time.
// Don't save if invalid, because we want the syntax error every program start.
vfs->CreateFile(xmbPath, writeBuffer.Data(), writeBuffer.Size());
vfs->CreateFile(xmbPath, m_Data.m_Buffer, m_Data.m_Size);
m_XMBBuffer = writeBuffer.Data(); // add a reference
// Set up the XMBFile
const bool ok = Initialise((const char*)m_XMBBuffer.get());
// Set up the XMBData
const bool ok = Initialise(m_Data);
ENSURE(ok);
return PSRETURN_OK;
}
bool CXeromyces::ReadXMBFile(const PIVFS& vfs, const VfsPath& filename)
{
size_t size;
if(vfs->LoadFile(filename, m_XMBBuffer, size) < 0)
return false;
// if the game crashes during loading, (e.g. due to driver bugs),
// it sometimes leaves empty XMB files in the cache.
// reporting failure will cause our caller to re-generate the XMB.
if(size == 0)
return false;
ENSURE(size >= 4); // make sure it's at least got the initial header
// Set up the XMBFile
if(!Initialise((const char*)m_XMBBuffer.get()))
return false;
return true;
}
PSRETURN CXeromyces::LoadString(const char* xml, const std::string& validatorName /* = "" */)
{
ENSURE(g_XeromycesStarted);
@@ -242,185 +224,12 @@ PSRETURN CXeromyces::LoadString(const char* xml, const std::string& validatorNam
}
}
WriteBuffer writeBuffer;
CreateXMB(doc, writeBuffer);
m_Data.LoadXMLDoc(doc);
xmlFreeDoc(doc);
m_XMBBuffer = writeBuffer.Data(); // add a reference
// Set up the XMBFile
const bool ok = Initialise((const char*)m_XMBBuffer.get());
// Set up the XMBData
const bool ok = Initialise(m_Data);
ENSURE(ok);
return PSRETURN_OK;
}
static void FindNames(const xmlNodePtr node, std::set<std::string>& elementNames, std::set<std::string>& attributeNames)
{
elementNames.insert((const char*)node->name);
for (xmlAttrPtr attr = node->properties; attr; attr = attr->next)
attributeNames.insert((const char*)attr->name);
for (xmlNodePtr child = node->children; child; child = child->next)
if (child->type == XML_ELEMENT_NODE)
FindNames(child, elementNames, attributeNames);
}
static void OutputElement(const xmlNodePtr node, WriteBuffer& writeBuffer,
std::map<std::string, u32>& elementIDs,
std::map<std::string, u32>& attributeIDs
)
{
// Filled in later with the length of the element
size_t posLength = writeBuffer.Size();
writeBuffer.Append("????", 4);
writeBuffer.Append(&elementIDs[(const char*)node->name], 4);
u32 attrCount = 0;
for (xmlAttrPtr attr = node->properties; attr; attr = attr->next)
++attrCount;
writeBuffer.Append(&attrCount, 4);
u32 childCount = 0;
for (xmlNodePtr child = node->children; child; child = child->next)
if (child->type == XML_ELEMENT_NODE)
++childCount;
writeBuffer.Append(&childCount, 4);
// Filled in later with the offset to the list of child elements
size_t posChildrenOffset = writeBuffer.Size();
writeBuffer.Append("????", 4);
// Trim excess whitespace in the entity's text, while counting
// the number of newlines trimmed (so that JS error reporting
// can give the correct line number within the script)
std::string whitespace = " \t\r\n";
std::string text;
for (xmlNodePtr child = node->children; child; child = child->next)
{
if (child->type == XML_TEXT_NODE)
{
xmlChar* content = xmlNodeGetContent(child);
text += std::string((const char*)content);
xmlFree(content);
}
}
u32 linenum = xmlGetLineNo(node);
// Find the start of the non-whitespace section
size_t first = text.find_first_not_of(whitespace);
if (first == text.npos)
// Entirely whitespace - easy to handle
text = "";
else
{
// Count the number of \n being cut off,
// and add them to the line number
std::string trimmed (text.begin(), text.begin()+first);
linenum += std::count(trimmed.begin(), trimmed.end(), '\n');
// Find the end of the non-whitespace section,
// and trim off everything else
size_t last = text.find_last_not_of(whitespace);
text = text.substr(first, 1+last-first);
}
// Output text, prefixed by length in bytes
if (text.length() == 0)
{
// No text; don't write much
writeBuffer.Append("\0\0\0\0", 4);
}
else
{
// Write length and line number and null-terminated text
u32 nodeLen = u32(4 + text.length()+1);
writeBuffer.Append(&nodeLen, 4);
writeBuffer.Append(&linenum, 4);
writeBuffer.Append((void*)text.c_str(), nodeLen-4);
}
// Output attributes
for (xmlAttrPtr attr = node->properties; attr; attr = attr->next)
{
writeBuffer.Append(&attributeIDs[(const char*)attr->name], 4);
xmlChar* value = xmlNodeGetContent(attr->children);
u32 attrLen = u32(xmlStrlen(value)+1);
writeBuffer.Append(&attrLen, 4);
writeBuffer.Append((void*)value, attrLen);
xmlFree(value);
}
// Go back and fill in the child-element offset
u32 childrenOffset = (u32)(writeBuffer.Size() - (posChildrenOffset+4));
writeBuffer.Overwrite(&childrenOffset, 4, posChildrenOffset);
// Output all child elements
for (xmlNodePtr child = node->children; child; child = child->next)
if (child->type == XML_ELEMENT_NODE)
OutputElement(child, writeBuffer, elementIDs, attributeIDs);
// Go back and fill in the length
u32 length = (u32)(writeBuffer.Size() - posLength);
writeBuffer.Overwrite(&length, 4, posLength);
}
PSRETURN CXeromyces::CreateXMB(const xmlDocPtr doc, WriteBuffer& writeBuffer)
{
// Header
writeBuffer.Append(UnfinishedHeaderMagicStr, 4);
// Version
writeBuffer.Append(&XMBVersion, 4);
u32 i;
// Find the unique element/attribute names
std::set<std::string> elementNames;
std::set<std::string> attributeNames;
FindNames(xmlDocGetRootElement(doc), elementNames, attributeNames);
std::map<std::string, u32> elementIDs;
std::map<std::string, u32> attributeIDs;
// Output element names
i = 0;
u32 elementCount = (u32)elementNames.size();
writeBuffer.Append(&elementCount, 4);
for (const std::string& n : elementNames)
{
u32 textLen = (u32)n.length()+1;
writeBuffer.Append(&textLen, 4);
writeBuffer.Append((void*)n.c_str(), textLen);
elementIDs[n] = i++;
}
// Output attribute names
i = 0;
u32 attributeCount = (u32)attributeNames.size();
writeBuffer.Append(&attributeCount, 4);
for (const std::string& n : attributeNames)
{
u32 textLen = (u32)n.length()+1;
writeBuffer.Append(&textLen, 4);
writeBuffer.Append((void*)n.c_str(), textLen);
attributeIDs[n] = i++;
}
OutputElement(xmlDocGetRootElement(doc), writeBuffer, elementIDs, attributeIDs);
// file is now valid, so insert correct magic string
writeBuffer.Overwrite(HeaderMagicStr, 4, 0);
return PSRETURN_OK;
}