Project

General

Profile

1 53 aaronmk
#!/usr/bin/env python
2
# Maps one datasource to another, using a map spreadsheet if needed
3 986 aaronmk
# Exit status is the # of errors in the import, up to the maximum exit status
4 53 aaronmk
# For outputting an XML file to a PostgreSQL database, use the general format of
5
# http://vegbank.org/vegdocs/xml/vegbank_example_ver1.0.2.xml
6
7
import os.path
8
import sys
9 299 aaronmk
import xml.dom.minidom as minidom
10 53 aaronmk
11 266 aaronmk
sys.path.append(os.path.dirname(__file__)+"/../lib")
12 53 aaronmk
13 344 aaronmk
import exc
14 64 aaronmk
import opts
15 281 aaronmk
import Parser
16 982 aaronmk
import profiling
17 131 aaronmk
import sql
18 715 aaronmk
import strings
19 828 aaronmk
import term
20 310 aaronmk
import util
21 133 aaronmk
import xml_dom
22 86 aaronmk
import xml_func
23 53 aaronmk
24 84 aaronmk
def metadata_value(name):
25 892 aaronmk
    if util.is_str(name) and name.startswith(':'): return name[1:]
26 84 aaronmk
    else: return None
27
28 847 aaronmk
def main_():
29 131 aaronmk
    env_names = []
30
    def usage_err():
31 944 aaronmk
        raise SystemExit('Usage: '+opts.env_usage(env_names, True)+' '
32 847 aaronmk
            +sys.argv[0]+' [map_path...] [<input] [>output]')
33 838 aaronmk
34
    # Get db config from env vars
35
    db_config_names = ['engine', 'host', 'user', 'password', 'database']
36
    def get_db_config(prefix):
37
        return opts.get_env_vars(db_config_names, prefix, env_names)
38
    in_db_config = get_db_config('in')
39
    out_db_config = get_db_config('out')
40
    in_is_db = 'engine' in in_db_config
41
    out_is_db = 'engine' in out_db_config
42
43
    # Get other config from env vars
44
    end = util.cast(int, opts.get_env_var('n', None, env_names))
45
    start = util.cast(int, opts.get_env_var('start', '0', env_names))
46
    if end != None: end += start
47 946 aaronmk
    test = opts.env_flag('test', False, env_names)
48 947 aaronmk
    commit = opts.env_flag('commit', False, env_names) and not test
49 944 aaronmk
        # never commit in test mode
50 947 aaronmk
    redo = opts.env_flag('redo', test, env_names) and not commit
51
        # never redo in commit mode (manually run `make empty_db` instead)
52 946 aaronmk
    debug = opts.env_flag('debug', False, env_names)
53 859 aaronmk
    sql.run_raw_query.debug = debug
54 946 aaronmk
    verbose = debug or opts.env_flag('verbose', False, env_names)
55 944 aaronmk
    opts.get_env_var('profile_to', None, env_names) # add to env_names
56 131 aaronmk
57 838 aaronmk
    # Logging
58 662 aaronmk
    def log(msg, on=verbose):
59
        if on: sys.stderr.write(msg)
60 859 aaronmk
    def log_start(action, on=verbose): log(action+'...\n', on)
61 662 aaronmk
62 53 aaronmk
    # Parse args
63 510 aaronmk
    map_paths = sys.argv[1:]
64 512 aaronmk
    if map_paths == []:
65
        if in_is_db or not out_is_db: usage_err()
66
        else: map_paths = [None]
67 53 aaronmk
68 646 aaronmk
    def connect_db(db_config):
69 662 aaronmk
        log_start('Connecting to '+sql.db_config_str(db_config))
70 646 aaronmk
        db = sql.connect(db_config)
71
        return db
72
73 751 aaronmk
    out_is_xml_ref = [False]
74
75 838 aaronmk
    def process_input(root, row_ready, map_path):
76 512 aaronmk
        '''Inputs datasource to XML tree, mapping if needed'''
77
        # Load map header
78
        in_is_xpaths = True
79 751 aaronmk
        out_is_xpaths = True
80 512 aaronmk
        out_label = None
81
        if map_path != None:
82
            import copy
83
            import csv
84
85
            import xpath
86
87
            metadata = []
88
            mappings = []
89
            stream = open(map_path, 'rb')
90
            reader = csv.reader(stream)
91
            in_label, out_label = reader.next()[:2]
92
            def split_col_name(name):
93
                name, sep, root = name.partition(':')
94
                return name, sep != '', root
95
            in_label, in_is_xpaths, in_root = split_col_name(in_label)
96
            out_label, out_is_xpaths, out_root = split_col_name(out_label)
97
            has_types = out_root.startswith('/*s/') # outer elements are types
98
            for row in reader:
99
                in_, out = row[:2]
100
                if out != '':
101
                    if out_is_xpaths: out = xpath.parse(out_root+out)
102
                    mappings.append((in_, out))
103
            stream.close()
104
105
            root.ownerDocument.documentElement.tagName = out_label
106
        in_is_xml = in_is_xpaths and not in_is_db
107 751 aaronmk
        out_is_xml_ref[0] = out_is_xpaths and not out_is_db
108 56 aaronmk
109 512 aaronmk
        if in_is_xml:
110
            doc0 = minidom.parse(sys.stdin)
111
            if out_label == None: out_label = doc0.documentElement.tagName
112 53 aaronmk
113 838 aaronmk
        def process_rows(process_row, rows):
114
            '''Processes input rows
115
            @param process_row(in_row, i)
116 297 aaronmk
            '''
117 838 aaronmk
            i = -1 # in case for loop does not execute
118 314 aaronmk
            for i, row in enumerate(rows):
119 838 aaronmk
                if i < start: continue
120
                if end != None and i >= end: break
121
                process_row(row, i)
122
                row_ready(i, row)
123 978 aaronmk
            row_ct = i-start+1
124 982 aaronmk
            return row_ct
125 838 aaronmk
126
        def map_rows(get_value, rows):
127
            '''Maps input rows
128
            @param get_value(in_, row):str
129
            '''
130
            def process_row(row, i):
131 316 aaronmk
                row_id = str(i)
132
                for in_, out in mappings:
133
                    value = metadata_value(in_)
134 662 aaronmk
                    if value == None:
135 857 aaronmk
                        log_start('Getting '+str(in_), debug)
136 662 aaronmk
                        value = get_value(in_, row)
137 715 aaronmk
                    if value != None: xpath.put_obj(root, out, row_id,
138
                        has_types, strings.cleanup(value))
139 982 aaronmk
            return process_rows(process_row, rows)
140 297 aaronmk
141 310 aaronmk
        if map_path == None:
142
            iter_ = xml_dom.NodeElemIter(doc0.documentElement)
143
            util.skip(iter_, xml_dom.is_text) # skip metadata
144 982 aaronmk
            row_ct = process_rows(lambda row, i: root.appendChild(row), iter_)
145 309 aaronmk
        elif in_is_db:
146 130 aaronmk
            assert in_is_xpaths
147 126 aaronmk
148 117 aaronmk
            import db_xml
149
150 161 aaronmk
            in_root_xml = xpath.path2xml(in_root)
151 164 aaronmk
            for i, mapping in enumerate(mappings):
152
                in_, out = mapping
153
                if metadata_value(in_) == None:
154 168 aaronmk
                    mappings[i] = (xpath.path2xml(in_root+'/'+in_), out)
155 126 aaronmk
156 646 aaronmk
            in_db = connect_db(in_db_config)
157 133 aaronmk
            in_pkeys = {}
158 297 aaronmk
            def get_value(in_, row):
159 167 aaronmk
                pkey, = row
160 297 aaronmk
                in_ = in_.cloneNode(True) # don't modify orig value!
161 886 aaronmk
                xml_dom.set_id(xpath.get(in_, in_root)[0], pkey)
162 297 aaronmk
                value = sql.value_or_none(db_xml.get(in_db, in_, in_pkeys))
163
                if value != None: return str(value)
164
                else: return None
165 982 aaronmk
            row_ct = map_rows(get_value, sql.rows(db_xml.get(in_db, in_root_xml,
166 865 aaronmk
                in_pkeys, end, 0)))
167 117 aaronmk
            in_db.close()
168 161 aaronmk
        elif in_is_xml:
169 297 aaronmk
            def get_value(in_, row):
170 886 aaronmk
                nodes = xpath.get(row, in_)
171
                if nodes != []: return xml_dom.value(nodes[0])
172 297 aaronmk
                else: return None
173 890 aaronmk
            rows = xpath.get(doc0.documentElement, in_root, limit=end)
174 886 aaronmk
            if rows == []: raise SystemExit('Map error: Root "'+in_root
175 883 aaronmk
                +'" not found in input')
176 982 aaronmk
            row_ct = map_rows(get_value, rows)
177 56 aaronmk
        else: # input is CSV
178 133 aaronmk
            map_ = dict(mappings)
179 59 aaronmk
            reader = csv.reader(sys.stdin)
180 84 aaronmk
            cols = reader.next()
181 162 aaronmk
            col_idxs = dict([(value, idx) for idx, value in enumerate(cols)])
182 164 aaronmk
            for i, mapping in enumerate(mappings):
183
                in_, out = mapping
184
                if metadata_value(in_) == None:
185
                    try: mappings[i] = (col_idxs[in_], out)
186
                    except KeyError: pass
187 162 aaronmk
188 297 aaronmk
            def get_value(in_, row):
189
                value = row[in_]
190
                if value != '': return value
191
                else: return None
192 982 aaronmk
            row_ct = map_rows(get_value, reader)
193
194
        return row_ct
195 53 aaronmk
196 838 aaronmk
    def process_inputs(root, row_ready):
197 982 aaronmk
        row_ct = 0
198
        for map_path in map_paths:
199
            row_ct += process_input(root, row_ready, map_path)
200
        return row_ct
201 512 aaronmk
202 984 aaronmk
    ex_tracker = exc.ExPercentTracker(iter_text='row')
203 982 aaronmk
    profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
204 985 aaronmk
205 512 aaronmk
    doc = xml_dom.create_doc()
206 316 aaronmk
    root = doc.documentElement
207 130 aaronmk
    if out_is_db:
208 53 aaronmk
        import db_xml
209
210 646 aaronmk
        out_db = connect_db(out_db_config)
211 310 aaronmk
        out_pkeys = {}
212 53 aaronmk
        try:
213 947 aaronmk
            if redo: sql.empty_db(out_db)
214 982 aaronmk
            row_ins_ct_ref = [0]
215 449 aaronmk
216 838 aaronmk
            def row_ready(row_num, input_row):
217 452 aaronmk
                def on_error(e):
218 835 aaronmk
                    exc.add_msg(e, term.emph('row #:')+' '+str(row_num))
219 828 aaronmk
                    exc.add_msg(e, term.emph('input row:')+'\n'+str(input_row))
220
                    exc.add_msg(e, term.emph('output row:')+'\n'+str(root))
221 452 aaronmk
                    ex_tracker.track(e)
222
223 449 aaronmk
                xml_func.process(root, on_error)
224 442 aaronmk
                if not xml_dom.is_empty(root):
225
                    assert xml_dom.has_one_child(root)
226
                    try:
227 982 aaronmk
                        sql.with_savepoint(out_db,
228
                            lambda: db_xml.put(out_db, root.firstChild,
229
                                out_pkeys, row_ins_ct_ref, on_error))
230 442 aaronmk
                        if commit: out_db.commit()
231 449 aaronmk
                    except sql.DatabaseErrors, e: on_error(e)
232 316 aaronmk
                root.clear()
233 449 aaronmk
234 982 aaronmk
            row_ct = process_inputs(root, row_ready)
235
            sys.stdout.write('Inserted '+str(row_ins_ct_ref[0])+
236 460 aaronmk
                ' new rows into database\n')
237 53 aaronmk
        finally:
238 133 aaronmk
            out_db.rollback()
239
            out_db.close()
240 751 aaronmk
    else:
241 759 aaronmk
        def on_error(e): ex_tracker.track(e)
242 838 aaronmk
        def row_ready(row_num, input_row): pass
243 982 aaronmk
        row_ct = process_inputs(root, row_ready)
244 759 aaronmk
        xml_func.process(root, on_error)
245 751 aaronmk
        if out_is_xml_ref[0]:
246
            doc.writexml(sys.stdout, **xml_dom.prettyxml_config)
247
        else: # output is CSV
248
            raise NotImplementedError('CSV output not supported yet')
249 985 aaronmk
250 982 aaronmk
    profiler.stop(row_ct)
251
    ex_tracker.add_iters(row_ct)
252 990 aaronmk
    if verbose:
253
        sys.stderr.write('Processed '+str(row_ct)+' input rows\n')
254
        sys.stderr.write(profiler.msg()+'\n')
255
        sys.stderr.write(ex_tracker.msg()+'\n')
256 985 aaronmk
    ex_tracker.exit()
257 53 aaronmk
258 847 aaronmk
def main():
259
    try: main_()
260
    except Parser.SyntaxException, e: raise SystemExit(str(e))
261
262 846 aaronmk
if __name__ == '__main__':
263 847 aaronmk
    profile_to = opts.get_env_var('profile_to', None)
264
    if profile_to != None:
265
        import cProfile
266
        cProfile.run(main.func_code, profile_to)
267
    else: main()