Project

General

Profile

1 53 aaronmk
#!/usr/bin/env python
2
# Maps one datasource to another, using a map spreadsheet if needed
3
# For outputting an XML file to a PostgreSQL database, use the general format of
4
# http://vegbank.org/vegdocs/xml/vegbank_example_ver1.0.2.xml
5
6
import os.path
7
import sys
8 299 aaronmk
import xml.dom.minidom as minidom
9 53 aaronmk
10 266 aaronmk
sys.path.append(os.path.dirname(__file__)+"/../lib")
11 53 aaronmk
12 344 aaronmk
import exc
13 64 aaronmk
import opts
14 281 aaronmk
import Parser
15 131 aaronmk
import sql
16 715 aaronmk
import strings
17 828 aaronmk
import term
18 310 aaronmk
import util
19 133 aaronmk
import xml_dom
20 86 aaronmk
import xml_func
21 53 aaronmk
22 84 aaronmk
def metadata_value(name):
23 892 aaronmk
    if util.is_str(name) and name.startswith(':'): return name[1:]
24 84 aaronmk
    else: return None
25
26 847 aaronmk
def main_():
27 434 aaronmk
    ex_tracker = exc.ExTracker()
28
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
            sys.stderr.write('Processed '+str(i-start+1)+' input rows\n')
124
125
        def map_rows(get_value, rows):
126
            '''Maps input rows
127
            @param get_value(in_, row):str
128
            '''
129
            def process_row(row, i):
130 316 aaronmk
                row_id = str(i)
131
                for in_, out in mappings:
132
                    value = metadata_value(in_)
133 662 aaronmk
                    if value == None:
134 857 aaronmk
                        log_start('Getting '+str(in_), debug)
135 662 aaronmk
                        value = get_value(in_, row)
136 715 aaronmk
                    if value != None: xpath.put_obj(root, out, row_id,
137
                        has_types, strings.cleanup(value))
138 838 aaronmk
            process_rows(process_row, rows)
139 297 aaronmk
140 310 aaronmk
        if map_path == None:
141
            iter_ = xml_dom.NodeElemIter(doc0.documentElement)
142
            util.skip(iter_, xml_dom.is_text) # skip metadata
143 838 aaronmk
            process_rows(lambda row, i: root.appendChild(row), iter_)
144 309 aaronmk
        elif in_is_db:
145 130 aaronmk
            assert in_is_xpaths
146 126 aaronmk
147 117 aaronmk
            import db_xml
148
149 161 aaronmk
            in_root_xml = xpath.path2xml(in_root)
150 164 aaronmk
            for i, mapping in enumerate(mappings):
151
                in_, out = mapping
152
                if metadata_value(in_) == None:
153 168 aaronmk
                    mappings[i] = (xpath.path2xml(in_root+'/'+in_), out)
154 126 aaronmk
155 646 aaronmk
            in_db = connect_db(in_db_config)
156 133 aaronmk
            in_pkeys = {}
157 297 aaronmk
            def get_value(in_, row):
158 167 aaronmk
                pkey, = row
159 297 aaronmk
                in_ = in_.cloneNode(True) # don't modify orig value!
160 886 aaronmk
                xml_dom.set_id(xpath.get(in_, in_root)[0], pkey)
161 297 aaronmk
                value = sql.value_or_none(db_xml.get(in_db, in_, in_pkeys))
162
                if value != None: return str(value)
163
                else: return None
164 838 aaronmk
            map_rows(get_value, sql.rows(db_xml.get(in_db, in_root_xml,
165 865 aaronmk
                in_pkeys, end, 0)))
166 117 aaronmk
            in_db.close()
167 161 aaronmk
        elif in_is_xml:
168 297 aaronmk
            def get_value(in_, row):
169 886 aaronmk
                nodes = xpath.get(row, in_)
170
                if nodes != []: return xml_dom.value(nodes[0])
171 297 aaronmk
                else: return None
172 890 aaronmk
            rows = xpath.get(doc0.documentElement, in_root, limit=end)
173 886 aaronmk
            if rows == []: raise SystemExit('Map error: Root "'+in_root
174 883 aaronmk
                +'" not found in input')
175 890 aaronmk
            map_rows(get_value, rows)
176 56 aaronmk
        else: # input is CSV
177 133 aaronmk
            map_ = dict(mappings)
178 59 aaronmk
            reader = csv.reader(sys.stdin)
179 84 aaronmk
            cols = reader.next()
180 162 aaronmk
            col_idxs = dict([(value, idx) for idx, value in enumerate(cols)])
181 164 aaronmk
            for i, mapping in enumerate(mappings):
182
                in_, out = mapping
183
                if metadata_value(in_) == None:
184
                    try: mappings[i] = (col_idxs[in_], out)
185
                    except KeyError: pass
186 162 aaronmk
187 297 aaronmk
            def get_value(in_, row):
188
                value = row[in_]
189
                if value != '': return value
190
                else: return None
191 838 aaronmk
            map_rows(get_value, reader)
192 53 aaronmk
193 838 aaronmk
    def process_inputs(root, row_ready):
194
        for map_path in map_paths: process_input(root, row_ready, map_path)
195 512 aaronmk
196 53 aaronmk
    # Output XML tree
197 512 aaronmk
    doc = xml_dom.create_doc()
198 316 aaronmk
    root = doc.documentElement
199 130 aaronmk
    if out_is_db:
200 53 aaronmk
        import db_xml
201
202 646 aaronmk
        out_db = connect_db(out_db_config)
203 310 aaronmk
        out_pkeys = {}
204 53 aaronmk
        try:
205 947 aaronmk
            if redo: sql.empty_db(out_db)
206 53 aaronmk
            row_ct_ref = [0]
207 449 aaronmk
208 838 aaronmk
            def row_ready(row_num, input_row):
209 452 aaronmk
                def on_error(e):
210 835 aaronmk
                    exc.add_msg(e, term.emph('row #:')+' '+str(row_num))
211 828 aaronmk
                    exc.add_msg(e, term.emph('input row:')+'\n'+str(input_row))
212
                    exc.add_msg(e, term.emph('output row:')+'\n'+str(root))
213 452 aaronmk
                    ex_tracker.track(e)
214
215 449 aaronmk
                xml_func.process(root, on_error)
216 442 aaronmk
                if not xml_dom.is_empty(root):
217
                    assert xml_dom.has_one_child(root)
218
                    try:
219
                        sql.with_savepoint(out_db, lambda: db_xml.put(out_db,
220 462 aaronmk
                            root.firstChild, out_pkeys, row_ct_ref, on_error))
221 442 aaronmk
                        if commit: out_db.commit()
222 449 aaronmk
                    except sql.DatabaseErrors, e: on_error(e)
223 316 aaronmk
                root.clear()
224 449 aaronmk
225 838 aaronmk
            process_inputs(root, row_ready)
226 717 aaronmk
            sys.stdout.write('Inserted '+str(row_ct_ref[0])+
227 460 aaronmk
                ' new rows into database\n')
228 53 aaronmk
        finally:
229 133 aaronmk
            out_db.rollback()
230
            out_db.close()
231 751 aaronmk
    else:
232 759 aaronmk
        def on_error(e): ex_tracker.track(e)
233 838 aaronmk
        def row_ready(row_num, input_row): pass
234
        process_inputs(root, row_ready)
235 759 aaronmk
        xml_func.process(root, on_error)
236 751 aaronmk
        if out_is_xml_ref[0]:
237
            doc.writexml(sys.stdout, **xml_dom.prettyxml_config)
238
        else: # output is CSV
239
            raise NotImplementedError('CSV output not supported yet')
240 53 aaronmk
241 847 aaronmk
def main():
242
    try: main_()
243
    except Parser.SyntaxException, e: raise SystemExit(str(e))
244
245 846 aaronmk
if __name__ == '__main__':
246 847 aaronmk
    profile_to = opts.get_env_var('profile_to', None)
247
    if profile_to != None:
248
        import cProfile
249
        cProfile.run(main.func_code, profile_to)
250
    else: main()