From e09a08bd00274c9e974137d490733cd834b5c662 Mon Sep 17 00:00:00 2001
From: Matthew Honnibal <honnibal@gmail.com>
Date: Mon, 1 Jun 2015 23:06:30 +0200
Subject: [PATCH] * Add copy_state function

---
 spacy/syntax/_state.pxd |  3 ++-
 spacy/syntax/_state.pyx | 29 +++++++++++++++++++++++++++++
 2 files changed, 31 insertions(+), 1 deletion(-)

diff --git a/spacy/syntax/_state.pxd b/spacy/syntax/_state.pxd
index 5ffc1f063..ee89d3d59 100644
--- a/spacy/syntax/_state.pxd
+++ b/spacy/syntax/_state.pxd
@@ -106,7 +106,8 @@ cdef int head_in_buffer(const State *s, const int child, const int* gold) except
 cdef int children_in_stack(const State *s, const int head, const int* gold) except -1
 cdef int head_in_stack(const State *s, const int child, const int* gold) except -1
 
-cdef State* new_state(Pool mem, TokenC* sent, const int sent_length) except NULL
+cdef State* new_state(Pool mem, const TokenC* sent, const int sent_length) except NULL
+cdef int copy_state(State* dest, const State* src) except -1
 
 cdef int count_left_kids(const TokenC* head) nogil
 
diff --git a/spacy/syntax/_state.pyx b/spacy/syntax/_state.pyx
index 3aae85773..74167319f 100644
--- a/spacy/syntax/_state.pyx
+++ b/spacy/syntax/_state.pyx
@@ -21,9 +21,17 @@ cdef int add_dep(State *s, int head, int child, int label) except -1:
         s.sent[head].r_kids |= 1 << (-dist)
         s.sent[head].r_edge = child - head
         # Walk up the tree, setting right edge
+        n_iter = 0
+        start = head
         while s.sent[head].head != 0:
             head += s.sent[head].head
             s.sent[head].r_edge = child - head
+            n_iter += 1
+            if n_iter >= s.sent_len:
+                tree = [(i + s.sent[i].head) for i in range(s.sent_len)]
+                msg = "Error adding dependency (%d, %d). Could not find root of tree: %s"
+                msg = msg % (start, child, tree)
+                raise Exception(msg)
     else:
         s.sent[head].l_kids |= 1 << dist
         s.sent[head].l_edge = (child + s.sent[child].l_edge) - head
@@ -155,6 +163,27 @@ cdef State* new_state(Pool mem, const TokenC* sent, const int sent_len) except N
     return s
 
 
+cdef int copy_state(State* dest, const State* src) except -1:
+    assert dest.sent_len == src.sent_len
+    # Copy stack --- remember stack uses pointer arithmetic, so stack[-stack_len]
+    # is the last word of the stack.
+    dest.stack += (src.stack_len - dest.stack_len)
+    for i in range(src.stack_len):
+        dest.stack[-i] = src.stack[-i]
+    dest.stack_len = src.stack_len 
+    # Copy sentence (i.e. the parse), up to and including word i.
+    memcpy(dest.sent, src.sent, sizeof(TokenC) * src.sent_len)
+    dest.i = src.i
+    # Copy assigned entities --- also pointer arithmetic
+    dest.ent += (src.ents_len - dest.ents_len)
+    for i in range(src.ents_len):
+        dest.ent[-i] = src.ent[-i]
+    dest.ents_len = src.ents_len
+    assert dest.sent[dest.i].head == src.sent[src.i].head
+    if dest.stack_len > 0:
+        assert dest.stack[0] < dest.i
+
+
 # From https://en.wikipedia.org/wiki/Hamming_weight
 cdef inline uint32_t _popcount(uint32_t x) nogil:
     """Find number of non-zero bits."""