From 8788d34bf56672ac0adb0dc7cc54034b2a95e5a5 Mon Sep 17 00:00:00 2001
From: Kenta Sato <bicycle1885@gmail.com>
Date: Mon, 28 Nov 2016 13:05:58 +0900
Subject: [PATCH] WIP: make find faster

---
 src/xpath.jl | 109 +++++++++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 101 insertions(+), 8 deletions(-)

diff --git a/src/xpath.jl b/src/xpath.jl
index d1b0013..68e4aa8 100644
--- a/src/xpath.jl
+++ b/src/xpath.jl
@@ -49,8 +49,7 @@ end
 Find the first node matching `xpath` XPath query from `doc`.
 """
 function Base.findfirst(doc::Document, xpath::AbstractString)
-    # string("(", xpath, ")[position()=1]") may be faster
-    return first(find(doc, xpath))
+    return findfirst(doc.node, xpath)
 end
 
 """
@@ -59,8 +58,7 @@ end
 Find the last node matching `xpath` XPath query from `doc`.
 """
 function Base.findlast(doc::Document, xpath::AbstractString)
-    # string("(", xpath, ")[position()=last()]") may be faster
-    return last(find(doc, xpath))
+    return findlast(doc.node, xpath)
 end
 
 """
@@ -71,6 +69,19 @@ Find nodes matching `xpath` XPath query starting from `node`.
 The `ns` argument is an iterator of namespace prefix and URI pairs.
 """
 function Base.find(node::Node, xpath::AbstractString, ns=namespaces(node))::Vector{Node}
+    if is_asterisk(xpath)
+        return elements(node)
+    elseif is_ncname(xpath)
+        node_ptr = first_element_ptr(node.ptr)
+        nodes = Node[]
+        while node_ptr != C_NULL
+            if is_named_as(node_ptr, xpath)
+                push!(nodes, Node(node_ptr))
+            end
+            node_ptr = next_element_ptr(node_ptr)
+        end
+        return nodes
+    end
     context_ptr = new_xpath_context(document(node))
     if context_ptr == C_NULL
         throw_xml_error()
@@ -104,8 +115,30 @@ end
 Find the first node matching `xpath` XPath query starting from `node`.
 """
 function Base.findfirst(node::Node, xpath::AbstractString, ns=namespaces(node))
-    # string("(", xpath, ")[position()=1]") may be faster
-    return first(find(node, xpath, ns))
+    if is_asterisk(xpath)  # any element
+        node_ptr = first_element_ptr(node.ptr)
+        if node_ptr == C_NULL
+            throw_no_matching_error()
+        end
+        return Node(node_ptr)
+    elseif is_ncname(xpath)  # without namespace prefix
+        node_ptr = first_element_ptr(node.ptr)
+        while node_ptr != C_NULL
+            if is_named_as(node_ptr, xpath)
+                return Node(node_ptr)
+            end
+            node_ptr = next_element_ptr(node_ptr)
+        end
+        throw_no_matching_error()
+    #elseif is_qname(xpath)  # with namespace prefix
+    # TODO
+    else
+        ret = find(node, xpath, ns)
+        if isempty(ret)
+            throw_no_matching_error()
+        end
+        return first(ret)
+    end
 end
 
 """
@@ -114,8 +147,34 @@ end
 Find the last node matching `xpath` XPath query starting from `node`.
 """
 function Base.findlast(node::Node, xpath::AbstractString, ns=namespaces(node))
-    # string("(", xpath, ")[position()=last()]") may be faster
-    return last(find(node, xpath, ns))
+    if is_asterisk(xpath)  # any element
+        node_ptr = last_element_ptr(node.ptr)
+        if node_ptr == C_NULL
+            throw_no_matching_error()
+        end
+        return Node(node_ptr)
+    elseif is_ncname(xpath)  # without namespace prefix
+        node_ptr = last_element_ptr(node.ptr)
+        while node_ptr != C_NULL
+            if is_named_as(node_ptr, xpath)
+                return Node(node_ptr)
+            end
+            node_ptr = prev_element_ptr(node_ptr)
+        end
+        throw_no_matching_error()
+    # elseif is_qname(xpath)  # with namespace prefix
+    # TODO
+    else
+        ret = find(node, xpath, ns)
+        if isempty(ret)
+            throw_no_matching_error()
+        end
+        return last(ret)
+    end
+end
+
+function throw_no_matching_error()
+    throw(ArgumentError("no matching nodes"))
 end
 
 function new_xpath_context(doc)
@@ -162,3 +221,37 @@ function free(ptr::Ptr{_XPathObject})
         (Ptr{Void},),
         ptr)
 end
+
+# Check if `node` is named as `name` (same as `name(node) == name` but faster).
+function is_named_as(node_ptr, name)
+    node_str = unsafe_load(node_ptr)
+    if node_str.name == C_NULL
+        return false
+    end
+    ret = ccall(
+        (:xmlStrEqual, libxml2),
+        Cint,
+        (Cstring, Cstring),
+        node_str.name, name)
+    return ret == 1
+end
+
+function is_asterisk(name)
+    return name == "*"
+end
+
+function is_ncname(name)
+    return ccall(
+        (:xmlValidateNCName, libxml2),
+        Cint,
+        (Cstring, Cint),
+        name, 0) == 0
+end
+
+function is_qname(name)
+    return ccall(
+        (:xmlValidateQName, libxml2),
+        Cint,
+        (Cstring, Cint),
+        name, 0) == 0
+end