def parseDesc(filename): descData = read(filename) descDict = {} for row in descData: try: descDict[row[0]] = row[1:] except Exception, e: print e
def __init__(self, pathname): import ConfigParser cf = ConfigParser.ConfigParser() cf.read(pathname) self.srcKeyPathname = cf.get("baseconf", "srcKeyPathname") self.recordPathname = cf.get("baseconf", "recordPathname") self.resultPathname = cf.get("baseconf", "resultPathname") self.driverPathname = cf.get("baseconf", "driverPathname") # self.keynames=cf.get("baseconf","keynames") if os.path.exists(self.recordPathname): self.recorddata = read(self.recordPathname) else: self.recorddata = [] if os.path.exists(self.resultPathname): self.result = read(self.resultPathname) else: self.result = [] self.srcKeys = read(self.srcKeyPathname)
@author: yangjie ''' from com.office.util.codeUtil import forbidCodeErr forbidCodeErr() from com.office.util.sqlUtil import save from com.office.util import fileUtil from com.office.template2 import DBNAME from com.office.util.excelUtil import read if __name__ == "__main__": keywords = [] srcpathname = r"/Users/yangjie/Downloads/qq/prescription_use_method" if srcpathname.endswith("xls") or srcpathname.endswith("xlsx"): tableData = read(srcpathname, "Sheet1") for row in tableData: keywords.append(row[0]) else: f1 = fileUtil.readFile(srcpathname) keywords = f1.split("\r\n") if len(keywords) < 2: keywords = f1.split("\r") if len(keywords) < 2: keywords = f1.split("\n") #上面的代码就是将原始数据转换城一个列表 data = [] save_sql = """insert into "main"."keys" ( "status", "keyword") values ( ?, ?);""" import sqlite3
#!/usr/bin/python # -*-coding:UTF-8-*- # encoding=utf8 from com.office.util.excelUtil import read tableData = read("/Users/yangjie/Desktop/juece2.xlsx") print tableData mainMap = {} judgeRed = tableData[0][1] judgeSize = tableData[0][2] tableData = tableData[1:] for rowData in tableData: name, hong, da, result = rowData print name, hong, da, result if not mainMap.has_key(judgeRed): mainMap[judgeRed] = {} tmpMap = mainMap[judgeRed] if not tmpMap.has_key(judgeSize): tmpMap[judgeSize] = {} tmpDaMap = tmpMap[judgeSize] tmpDaMap[da] = {"d": "d"} from juece import createPlot print mainMap createPlot(mainMap)
#!/usr/bin/python # -*-coding:UTF-8-*- # encoding=utf8 import matplotlib.pyplot as plt from com.office.util.excelUtil import read from com.office.util.plotutil.common.colorUtil import getRandomColor from com.office.util.plotutil.common.descUtil import parseDesc fig, axes = plt.subplots() all_data = read('scatterplot.xls') descDict = parseDesc('desc.xls') if descDict.has_key("xaxisgrid"): axes.xaxis.grid(True) if descDict.has_key("yaxisgrid"): axes.yaxis.grid(True) xlistList = [] ylistList = [] for i in range(len(all_data) / 2): xlistList.append(all_data[i]) ylistList.append(all_data[i + 1]) if descDict.has_key("colors"): colors = descDict["colors"] else: colors = getRandomColor(len(all_data)) barList = [] xticklist = all_data[0] typeList = [] for xlist, ylist, color in zip(xlistList, ylistList, colors): curType = plt.scatter(xlist, ylist, s=20, c=color) #s (点的大小)
for start in (0, 1): for x in np.linspace(0, 1, needed[start] + 2): colours.append((base[start] * (1.0 - x)) + (base[start + 1] * x)) return [pastel(c) for c in colours[0:n]] lens = 10 srcFiles = srcFiles[:lens] cs = get_colours(lens - 1) print cs for srcFile, c in zip(srcFiles, cs): x = [] y = [] print "parse", srcFile srcData = read(os.path.join("yaohao", srcFile)) src = [] d = {} for data in srcData[0]: datar = int(data) / 100000000000 if not d.has_key(datar): d[datar] = 0 d[datar] += 1 for key in d: y.append(d[key]) x.append(key) print c print "x", "y" print "x", "y"
cu.execute(select_sql) result = cu.fetchall() return result except Exception, e: del e finally: if conn: conn.close() if __name__ == "__main__": keywords = [] srcpathname = r"/Users/yangjie/mywork/workspace/mypython/com/office/template4ncbiplus/db/update_pubid.xlsx" src2pathname = r"/Users/yangjie/mywork/workspace/mypython/com/office/template4ncbiplus/db/update_pubid2.xlsx" destPathname = r"/Users/yangjie/mywork/workspace/mypython/com/office/template4ncbiplus/db/result1023_plus.xlsx" tableData = read(srcpathname, "Sheet2") table2Data = read(src2pathname, "Sheet2") resultTable = [] updateCount = 0 for row in tableData: for row2 in table2Data: if row[4] == row2[4]: row = row2 break if row[3] == u'Not Available': result = getFromDB(row[4]) if result: info = result[0][1:] row[3] = info[3] row[6] = info[2] row[7] = info[4]
#!/usr/bin/python # -*-coding:UTF-8-*- # encoding=utf8 import matplotlib.pyplot as plt from com.office.util.excelUtil import read from com.office.util.plotutil.common.colorUtil import getRandomColor from com.office.util.plotutil.common.descUtil import parseDesc fig, axes = plt.subplots() all_data = read('boxplot.xls') for i in range(len(all_data)): for y in range(len(all_data[i])): if "" == all_data[i][y]: all_data[i] = all_data[i][:y] break descDict = parseDesc('desc.xls') if descDict.has_key("xaxisgrid"): axes.xaxis.grid(True) if descDict.has_key("yaxisgrid"): axes.yaxis.grid(True) bplot = plt.boxplot( all_data, notch=False, # box instead of notch shape sym='rs', # red squares for outliers vert=True, patch_artist=True) # vertical box aligmnent #init color if descDict.has_key("colors"):
#!/usr/bin/python # -*-coding:UTF-8-*- # encoding=utf8 import numpy import pylab from com.office.util.excelUtil import read from matplotlib import rcParams pylab.figure(figsize=(9, 6), dpi=80, edgecolor='y') srcData = read("resulttiezi.xls") keylist = srcData[0][:6] vallist = srcData[1][:6] barwidth = 0.2 xVal = numpy.arange(len(keylist)) / 3.0 print xVal print xVal + barwidth pylab.xticks(xVal + barwidth + 0.1, keylist, rotation=25) pylab.bar(xVal + barwidth, vallist, width=barwidth, color='#FFFF00', edgecolor='#8B2252') pylab.title(u'测试分析图', loc="right") pylab.show()
#!/usr/bin/python #-*-coding:UTF-8-*- # encoding=utf8 from com.office.util.excelUtil import read, write srcDate = read("六神曲.xlsx") srcDate = srcDate[1:] cache = [] lastSrc = "" for rowDate in srcDate: if rowDate[2] != "": cacheline = [] cacheline.append(rowDate[2]) if rowDate[3] != "": lastSrc = rowDate[3] cacheline.append(lastSrc) cache.append(cacheline) resultDate = [] for rowDate in srcDate: newLine = [] newLine.append(rowDate[0]) if rowDate[1] != "": newLine.append(rowDate[1]) newLine.append("TCMID") else: if len(cache) != 0: takeCache = cache[0] cache.remove(takeCache) newLine.extend(takeCache) resultDate.append(newLine)
#!/usr/bin/python # -*-coding:UTF-8-*- # encoding=utf8 import numpy import pylab from com.office.util.excelUtil import read import os from com.office.util import fileUtil srcFiles = fileUtil.getFiles("../sandiantu/yaohao") d = {} for srcFile in srcFiles[:5]: srcData = read(os.path.join("../sandiantu/yaohao", srcFile)) src = [] print len(srcData[0]) for data in srcData[0]: datar = int(data) / 100000000000 if not d.has_key(datar): d[datar] = 0 d[datar] += 1 keylist = [] vallist = [] vallist2 = [] vallist3 = [] count = 10 fig, axes = pylab.subplots(nrows=1, ncols=2, figsize=(12, 5)) for key in d: if count < 0: break else:
#!/usr/bin/python #-*-coding:UTF-8-*- # encoding=utf8 from selenium import webdriver from com.office.util.codeUtil import forbidCodeErr from com.office.util.excelUtil import write, read from selenium.webdriver.common.keys import Keys from com.office.util.pubUtil import doInThread, retry forbidCodeErr() FILENAME_RECORD = "record.xls" FILENAME_RESULT = "result.xls" executable_path = "/Users/yangjie/Documents/env/jyang-site-packages/chromedriver_2.21/chromedriver_mac32/chromedriver" reusltData = [] try: reusltData = read(FILENAME_RESULT) except Exception, e: print e recoreDate = [] matchRecordData = [] try: recoreDate = read(FILENAME_RECORD) for row in recoreDate: matchRecordData.append(row[0]) except Exception, e: print e ''' retry方法作用是对可能出错的代码进行最多100次的重试 '''