Data_Importing.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Sun Dec 10 03:17:41 2017

@author: scott
"""

"""
This module is thought of as a set of high-level functions that can do
switching between file types and combining of multiple files.
TODO:
Low-level functions (many of them having to do with timestamp parsing) should
be placed in parsing_tools, so that they can be imported by any EC_MS module

Right now, the text_to_data function does a ton of stuff. Hopefully this will
not continue to be the case.

TODO: The goal is that file-type-specific things all get moved to modules named
for that specific file type or data producing equipment.


"""


import os
from pathlib import Path
import re
import codecs
import numpy as np

from .parsing_tools import (
    numerize,
    timestamp_match,
    get_creation_time,
    parse_date,
    timestring_to_epoch_time,
    epoch_time_to_timestamp,
)


def import_EC_data(
    full_path_name,
    title="get_from_file",
    data_type="EC",
    N_blank=10,
    verbose=True,
    header_string=None,
    timestamp=None,
):
    file_lines = import_text(full_path_name, verbose=verbose)
    dataset = text_to_data(
        file_lines,
        title="get_from_file",
        data_type="EC",
        N_blank=10,
        verbose=True,
        header_string=None,
        timestamp=None,
    )
    numerize(dataset)
    return dataset


"""
The following couple functions are adapted from EC_MS on 17L09
as last commited to EC_MS with code c1c6efa
They might benifit from a full rewrite, but not now.
"""


def import_text(full_path_name="current", verbose=True):
    """
    This method will import the full text of a file selected by user input as a
    list of lines.
    When I first wrote it for EC_MS, way back in the day, I made it so you can
    call it without any arguments, and then input. probably unecessary.
    """
    if verbose:
        print("\n\nfunction 'import_text' at your service!\n")

    if full_path_name == "input":
        full_path_name = input(
            "Enter full path for the file name as 'directory"
            + os.sep
            + "file.extension'"
        )
    if full_path_name == "current":
        full_path_name = os.getcwd()

    [directory_name, file_name] = os.path.split(full_path_name)
    if directory_name == "":
        directory_name = "."
    original_directory = os.getcwd()
    os.chdir(directory_name)

    if os.path.isdir(full_path_name) and not os.path.isfile(file_name):
        directory_name = full_path_name
        os.chdir(directory_name)
        ls_string = str(os.listdir())
        print("\n" + full_path_name + "\n ls: \n" + ls_string + "\n")
        file_name = input(
            "Directory given. Enter the full name of the file to import\n"
        )

    if verbose:
        print("directory: " + directory_name)
        print("importing data from " + file_name)

    possible_encodings = ["utf8", "iso8859_15"]
    # mpt files seem to be the latter encoding, even though they refer to themselves as ascii
    for encoding_type in possible_encodings:
        try:
            with codecs.open(file_name, "r", encoding=encoding_type) as file_object:
                file_lines = file_object.readlines()
            if verbose:
                print("Was able to readlines() with encoding " + encoding_type)
            break
        except UnicodeDecodeError:
            if verbose:
                print("Shit, some encoding problem in readlines() for " + encoding_type)
        except FileNotFoundError:
            print(
                "File not Found! file_name = "
                + file_name
                + "\nDirectory = "
                + os.getcwd()
                + "\nfiles = "
                + str(os.listdir(os.getcwd()))
            )
            raise
    else:
        print("couldn't read " + file_name + "\n ... may by due to an encoding issue")

    os.chdir(original_directory)

    if verbose:
        print("\nfunction 'import_text' finished!\n\n")
    return file_lines


def text_to_data(
    file_lines,
    title=None,
    timestamp=None,
    date="today",
    tstamp=None,
    tz=None,
    data_type="EC",
    sep=None,
    header_string=None,
    verbose=True,
):
    """
    This method will organize data in the lines of text from a file useful for
    electropy into a dictionary as follows (plus a few more keys)
    {'title':title, 'header':header, 'timestamp':timestamp,
     'data_cols':{colheader1, colheader2, ...},
     colheader1:data1, colheader2:data2, ...}

    Supported data types:
        'EC': text output (.mpt) from Biologic for any voltammetric technique
        'MS': text output from cinfdata for mass_time_scan, run by PyExpLabSys
        'SPEC': .csv file saved by SPEC diffraction program at SSRL BL2.1
        'XAS': text saved by XAS program at SSRL BL11.2
        'SI': text ouptut (.csv) from Zilien, Kenneth's software for Spectro Inlets
        'RGA': text output from Residual Gas Analysis program for mass spec
    """
    if verbose:
        print("\n\nfunction 'text_to_data' at your service!\n")

    # disect header
    N_lines = len(file_lines)  # number of header lines
    N_head = N_lines  # this will change when I find the line that tells me how ling the header is
    header_string = ""

    data = {}
    data["data_type"] = data_type
    data["timecols"] = {}

    commacols = []  # will catch if data is recorded with commas as decimals.

    loop = False

    if data_type == "SPEC" or data_type == "ULM":
        N_head = 1  # the column headers are the first line
        if sep is None:
            sep = ","
    elif data_type == "SI" or data_type == "MKS":  # Spectro Inlets data
        sep = "\t"  # despite it being a .csv, the separator is a tab.
    elif data_type == "RGA":
        sep = ","
        N_blank = 2
        data["channels"] = {}
    elif data_type == "CHI":  # CH Instruments potentiostat
        N_blank = 2
        sep = ","
    elif data_type == "MS":
        N_blank = 10

    if sep is None:  # EC and MS data all work with '\t'
        sep = "\t"

    # print("N_head = " + str(N_head))  # debugging

    n_blank = 0
    got_col_headers = False
    for nl, line in enumerate(file_lines):
        l = line.strip()
        if nl < N_head - 1:  # we're in the header

            if data_type == "EC":
                if title is None:
                    if re.search("File :", line):
                        title_object = re.search(r"[\S]*\Z", line.strip())
                        title = title_object.group()
                        if verbose:
                            print("name '" + title + "' found in line " + str(nl))
                if re.search(r"[Number ]*header lines", line):
                    N_head_object = re.search(r"[0-9][0-9]*", line)
                    N_head = int(N_head_object.group())
                    if verbose:
                        print("N_head '" + str(N_head) + "' found in line " + str(nl))
                elif timestamp is None and re.search("Acquisition started", line):
                    timestamp_object = re.search(timestamp_match, l)
                    timestamp = timestamp_object.group()
                    date = parse_date(l)
                    if verbose:
                        print("timestamp '" + timestamp + "' found in line " + str(nl))
                elif re.search("Number of loops", line):
                    # Then I want to add a loop number variable to data_cols
                    loop = True
                    data["loop number"] = []
                elif re.search("Loop", line):
                    n = int(re.search(r"^Loop \d+", line).group()[5:])
                    start = int(re.search(r"number \d+", line).group()[7:])
                    finish = int(re.search(r"to \d+", line).group()[3:])
                    N = finish - start + 1
                    data["loop number"] += N * [n]

            elif data_type == "MS":
                if len(line.strip()) == 0:
                    n_blank += 1
                    if n_blank >= N_blank and len(file_lines[nl + 1].strip()) > 0:
                        N_head = nl + 2
                    continue
                else:
                    n_blank = 0
                if title is None:
                    object1 = re.search(r'"Comment"[\s]*"[^"]*', line)
                    if object1:
                        string1 = object1.group()
                        title_object = re.search(r"[\S]*\Z", string1.strip())
                        title = title_object.group()[1:]
                        if verbose:
                            print("name '" + title + "' found in line " + str(nl))
                object2 = re.search(r'"Recorded at"[\s]*"[^"]*', line)
                if object2:
                    string2 = object2.group()
                    timestamp_object = re.search(timestamp_match, string2.strip())
                    timestamp = timestamp_object.group()
                    date = parse_date(l)
                    # ^convert yyyy-mm-dd to dd-mm-yyyy
                    if verbose:
                        print("timestamp '" + timestamp + "' found in line " + str(nl))

            elif data_type == "SI":  # Spectro Inlets data format
                items = [
                    item.strip() for item in line.split(sep) if len(item.strip()) > 0
                ]
                if nl < 10:
                    # print(items) # debugging
                    pass
                if len(items) == 0:
                    continue
                if title is None and items[0] == "name":
                    title = items[-1]
                    if verbose:
                        print("title '" + str(title) + "' found in line " + str(nl))
                if items[0] == "offset":
                    offset = float(items[-1])
                    data["SI offset"] = offset
                    if verbose:
                        print(
                            "SI offset '" + str(offset) + "' found in line " + str(nl)
                        )
                if items[0] == "data_start":
                    N_head = int(items[-1])
                    if verbose:
                        print("N_head '" + str(N_head) + "' found in line " + str(nl))
                if nl == N_head - 2:
                    col_preheaders = [item.strip() for item in line.split(sep)]
                    for i, preheader in enumerate(col_preheaders):
                        if len(preheader) == 0:
                            col_preheaders[i] = col_preheaders[i - 1]

            elif data_type == "MKS":  # old Spectro Inlets MS data format
                # this actually records absolute time (in a fucked up format),
                # so no need to read the timestring.
                if "[Scan Data" in line:  # then the data is coming.
                    N_head = nl + 2

            elif data_type == "RGA":
                if len(line.strip()) == 0:
                    n_blank += 1
                    if n_blank >= N_blank and len(file_lines[nl + 1].strip()) > 0:
                        N_head = nl + 2
                    continue
                else:
                    n_blank = 0
                if re.search("Start time", line):
                    tstamp, date, timestamp = timestring_to_epoch_time(
                        l, tz=tz, out="all", verbose=verbose
                    )
                if re.search(r"\A[0-9]+\s", l):  # lines starting with a number
                    items = [
                        item.strip()
                        for item in line.split(" ")
                        if len(item.strip()) > 0
                    ]
                    channel = "Channel#" + items[0]
                    mass = "M" + items[1].split(".")[0]
                    data["channels"][channel] = mass
                if "Analog Scan" in l:
                    col_headers = ["m/z", "signal/A"]
                    got_col_headers = True
                    print("got column headers! on line " + str(nl))

            elif data_type == "CHI":
                if len(line.strip()) == 0:
                    n_blank += 1
                    if n_blank >= N_blank and len(file_lines[nl + 1].strip()) > 0:
                        N_head = nl + 2
                    continue
                else:
                    n_blank = 0
                if nl == 0:  # record time (measurement finish time) on top line
                    if verbose:
                        print("finding tstamp from line: " + l)
                    tstamp, date, timestamp = timestring_to_epoch_time(
                        l, tz=tz, out="all", verbose=verbose
                    )
                if "Segment = " in line:
                    data["segments"] = line.split(" = ")[-1].strip()
                    last_segment_line = "Segment " + data["segments"] + ":"
                if "segments" in data and last_segment_line in line:
                    N_blank = 1
                if "Scan Rate (V/s)" in line:
                    data["scan rate"] = (
                        eval(line.split(" = ")[-1].strip()) * 1e3
                    )  # in mV/s
                if "Time/s" in line:  # then it's actually the column header line.
                    N_head = (
                        nl + 2
                    )  # the next line is a blank line, during which we handle the column headers
                    col_headers = l.split(sep)
                    got_col_headers = True  # to be used on next line (nl=N_head-1)

            header_string = header_string + line

        elif nl == N_head - 1:  # then it is the column-header line
            # (EC-lab includes the column-header line in header lines)
            # col_header_line = line
            if data_type == "RGA":  # there's no damn commas on the column header lines!
                if not got_col_headers:
                    col_headers = [
                        col.strip() for col in l.split(" ") if len(col.strip()) > 0
                    ]
            elif data_type == "CHI" and got_col_headers:
                pass
            else:
                col_headers = [col.strip() for col in l.split(sep=sep)]

            if data_type == "MKS":
                col_headers = [col.strip('"') for col in col_headers]

            if data_type == "SI":
                for i, col in enumerate(col_headers):
                    try:
                        col_headers[i] = col_preheaders[i] + " - " + col
                    except IndexError:
                        print("WARNING!!! Spectro Inlets pre-column header too short")
                        break

            print(f"col_headers = {col_headers}")  # debugging

            data["N_col"] = len(col_headers)
            data["data_cols"] = set(
                col_headers.copy()
            )  # will store names of columns containing data
            if not len(col_headers) == len(data["data_cols"]):
                print("WARNING: repeated column headers!!!")
                print("col_headers = " + str(col_headers))

            data["col_types"] = dict([(col, data_type) for col in col_headers])

            for col in col_headers:
                data[col] = []  # data will go here
            header_string = header_string + line  # include this line in the header
            if verbose:
                print("Data starting on line " + str(N_head) + "\n")

        elif len(l) == 0:
            # rga and chi text files skip a line after the column header
            continue

        else:  # data, baby!
            line_data = [dat.strip() for dat in l.split(sep=sep)]

            if data_type == "MKS":
                timestring = line_data[0].replace(".", ":")[1:-1]
                if "Annotations" in timestring:  # last line actually doesn't have data
                    continue
                yyyy, dd, mm = timestring[6:10], timestring[0:2], timestring[3:5]
                timestring = yyyy + "/" + mm + "/" + dd + timestring[-9:]
                t = timestring_to_epoch_time(timestring, verbose=False)
                line_data[0] = str(t)

            if not len(line_data) == len(col_headers):
                # print('Mismatch between col_headers and data on line ' + str(nl) + ' of ' + title) #debugging
                pass
            for col, x in zip(col_headers, line_data):
                if not col in data["data_cols"]:
                    # don't try adding data to a column that has already been determined not to have data!
                    continue
                try:
                    x = float(x)
                except ValueError:
                    if x == "":
                        continue  # added 17C22 to deal with data acquisition crashes.
                    try:
                        if verbose and not col in commacols:
                            print(
                                f"ValueError on value {x} in column {col} line {nl}\n"
                                + "Checking if yo're using commas as decimals in that column..."
                            )
                        x = x.replace(".", "")
                        # ^ in case there's also '.' as thousands separator, just get rid of it.
                        x = x.replace(",", ".")  # put '.' as decimals
                        x = float(x)
                    except ValueError:
                        if verbose:
                            print(list(zip(col_headers, line_data)))
                            print(
                                f"{title} in text_to_data: \nRemoved {col} from data columns"
                                + f" because ofvalue '{x}' at line {nl}\n"
                            )
                        data["data_cols"].remove(col)
                    else:
                        if not col in commacols:
                            if verbose:
                                print("... and you were, dumbass. I" "ll fix it.")
                            commacols += [col]
                data[col].append(x)

    if loop:
        data["data_cols"].add("loop number")
    data["title"] = title
    data["header"] = header_string
    data["timestamp"] = timestamp
    data["date"] = date
    if tstamp is None:
        tstamp = timestring_to_epoch_time(
            timestamp, date, tz=tz, verbose=verbose, out="tstamp"
        )

    if data_type == "EC":
        # rename potential and current variables,
        # so that synchronize can combine current data from different EC-lab techniques:
        if "<I>/mA" in data["data_cols"] and "I/mA" not in data["data_cols"]:
            # so that synchronize can combine current data from different EC-lab techniques
            data["data_cols"].add("I/mA")
            data["I/mA"] = data["<I>/mA"].copy()
        if "<Ewe>/V" in data["data_cols"] and "Ewe/V" not in data["data_cols"]:
            data["data_cols"].add("Ewe/V")
            data["Ewe/V"] = data["<Ewe>/V"].copy()
        # and populate timecols
        for col in data["data_cols"]:
            data["timecols"][col] = "time/s"
        data["t_str"] = "time/s"
    elif data_type == "MS":
        data["t_str"] = "<mass>-x"
    elif data_type == "RGA":
        for col in data["data_cols"]:
            data["timecols"][col] = "Time (s)"
        data["t_str"] = "Time (s)"
    elif data_type == "MKS":
        print(data.keys())  # debugging
        tstamp = data["Time"][0]
        data["Time"] = np.array(data["Time"]) - tstamp
        timestamp = epoch_time_to_timestamp(tstamp)
        date = None
        for col in data["data_cols"]:
            data["timecols"][col] = "Time"
        data["t_str"] = "Time"
    # I kind of think timecols should be defined for everything here, but it might not be

    data["timezone"] = tz
    data["tstamp"] = tstamp  # UNIX epoch time, for proper synchronization! :D

    if verbose:
        print("\nfunction 'text_to_data' finished!\n\n")
    return data


def import_data(*args, **kwargs):
    print(
        "'import_data' is now called 'load_from_file'!\n"
        + "Remember that next time, goof."
    )
    return load_from_file(*args, **kwargs)


def load_from_file(
    full_path_name="current",
    title="file",
    tstamp=None,
    timestamp=None,
    data_type="EC",
    tz=None,
    name=None,
    verbose=True,
):
    """
    This method will organize the data in a file useful for
    electropy into a dictionary as follows (plus a few more keys)
    {'title':title, 'header':header, 'timestamp':timestamp,
    'data_cols':[colheader1, colheader2, ...],
    colheader1:[data1], colheader2:[data2]...}
    """
    if verbose:
        print("\n\nfunction 'load_from_file' at your service!\n")

    if title == "file":
        folder, title = os.path.split(full_path_name)
    if folder == "":
        folder = "."
    if data_type == "PVMS":  # I want eventually to have one of these for everything
        from PVMassSpec import read_PVMS

        data = read_PVMS(full_path_name)
    else:
        file_lines = import_text(full_path_name, verbose)

        try:
            data = text_to_data(  # I want to split up the text_to_data function
                file_lines=file_lines,
                title=title,
                data_type=data_type,
                timestamp=timestamp,
                tz=tz,
                tstamp=tstamp,
                verbose=verbose,
            )
        except Exception as e:
            print(
                f"COULD NOT PARSE {full_path_name}!!! Got error = {e}. "
                "Returning an empty dictionary."
            )
            data = {"data_type": data_type, "title": "empty"}
            return data

    if tstamp is not None:  # then it overrides whatever text_to_data came up with.
        data["tstamp"] = tstamp
    elif data["tstamp"] is None:
        print(f"WARNING: no tstamp found in {full_path_name}. Looking in file name.")
        tstamp = timestring_to_epoch_time(full_path_name)
        if tstamp is None:
            print(
                "WARNING: no tstamp found in "
                + full_path_name
                + " file name either. Using file creation time."
            )
            tstamp = get_creation_time(full_path_name, verbose=verbose)
        data["tstamp"] = tstamp
    if "data_cols" not in data or len(data["data_cols"]) == 0:
        print("WARNING! empty dataset")
        data["empty"] = True
    else:
        numerize(data)
        data["empty"] = False
    if name is None:
        name = data["title"]
    data["name"] = name

    if data["empty"]:
        print("WARNING! load_from_file is returning an empty dataset")
    elif data_type == "SI":
        from .Combining import rename_SI_cols

        print("RENAMING SI COLS!")  # debugging
        rename_SI_cols(data)
    elif data_type == "RGA":
        from .Combining import rename_RGA_cols

        rename_RGA_cols(data)
    elif data_type == "CHI":
        from .Combining import parse_CHI_header, rename_CHI_cols, timeshift

        parse_CHI_header(data)
        rename_CHI_cols(data)
        dt = data["time/s"][-1] - data["time/s"][0]
        timeshift(data, dt)
    elif data_type == "ULM":
        from .Combining import rename_ULM_cols

        rename_ULM_cols(data)
    elif data_type == "PVMS":
        from .PVMassSpec import rename_PVMS_cols

        rename_PVMS_cols(data)
    elif data_type == "MKS":
        from .Combining import rename_MKS_cols

        rename_MKS_cols(data)

    if verbose:
        print("\nfunction 'load_from_file' finished!\n\n")
    return data


def load_EC_set(
    directory,
    EC_files=None,
    tag="01",
    suffix=None,
    data_type="EC",
    verbose=True,
    tz=None,
    exclude=[],
    fix_CP=False,
):
    """
    inputs:
        directory - path to folder containing your data, string

        EC_file - list of EC_files, list
            OR
        tag - shared start of EC files you want to load and combine, str AND
        suffix - ending of files, by default .mpt

        data_type - type of EC data. By default 'EC', meaning Biologic EC-Lab files
        tz - timezone, usually not needed
        verbose - makes the function talk to you.
    output
        EC_data - a dataset with the data from all specified EC files combined
            and sorted based on time. Additional columns loop_number and
            file_number are added to the dataset if relevant.
    """
    if verbose:
        print("\n\nfunction 'load_EC_set' at your service!\n")
    from .Combining import synchronize, sort_time

    if suffix is None:
        if data_type == "EC":
            suffix = ".mpt"
        elif data_type == "CHI":
            suffix = ".txt"

    lslist = os.listdir(directory)

    if EC_files is None:
        EC_files = [f for f in lslist if f.startswith(tag) and f.endswith(suffix)]
        if type(exclude) is str:
            exclude = [exclude]
        for excl in exclude:
            EC_files = [f for f in EC_files if not excl in f]
    elif type(EC_files) is str:
        EC_files = [EC_files]

    print(f"lslist = {lslist}\nEC_files = {EC_files}")  # debugging

    EC_datas = []
    for f in EC_files:
        try:
            data = load_from_file(
                Path(directory) / f, data_type=data_type, tz=tz, verbose=verbose
            )
        except OSError:
            print("WARNING: problem importing " + f + ". Continuing.")
            continue
        if fix_CP and "CP" in f:
            try:
                data["Ewe/V"] = data["Ewe-Ece/V"] + data["<Ece>/V"]
            except KeyError:
                print(
                    "WARNING! Could not fix CP for "
                    + f
                    + " because missing "
                    + " either Ece/V or Ewe-Ece/V"
                )
        EC_datas += [data]

    EC_data = synchronize(EC_datas, verbose=verbose, append=True, t_zero="first", tz=tz)
    if "loop number" in EC_data["data_cols"]:
        sort_time(EC_data, verbose=verbose)  # note, sort_time no longer returns!

    if verbose:
        print("\nfunction 'load_EC_set' finished!\n\n")
    return EC_data


def import_EC_set(*args, **kwargs):
    """
    See EC_MS.load_EC_set
    """
    print("import_EC_set has been renamed load_EC_set")
    return load_EC_set(*args, **kwargs)


def download_cinfdata_set(
    setup="sniffer", group_id=None, grouping_column=None, **kwargs
):

    if grouping_column is None:
        grouping_column, group_id = kwargs.popitem()

    from .Combining import synchronize

    try:
        from cinfdata import Cinfdata
    except ImportError:
        print(
            "the cinfdata module must be on your python path. It's here: \n"
            + "https://github.com/CINF/cinf_database/blob/master/cinfdata.py"
        )

    try:
        cinfd = Cinfdata(
            setup,
            grouping_column=grouping_column,
            allow_wildcards=True,
            label_column="mass_label",
        )
    except:
        raise  # untill I know exactly which error I'm trying to catch.
        print("couldn't connect. You should run gstm")
        # os.system('gstm')
        raise RuntimeError("Couldn't connect to cinfdata!")

    # obj = cinfd.get_metadata_group('2018-03-30 14:13:17')

    # all_datasets = cinfd.get_metadata_group('%')
    # the_list = [(ID, d['time'], d['comment']) for ID, d in all_datasets.items()]
    # print(the_list)

    obj = cinfd.get_metadata_group(group_id)
    # print(str(obj)) #

    idlists = {}  # keys will be time as string. values will be corresponding id's

    for key, value in obj.items():
        # label = value['mass_label']
        # print(label)
        timestamp = str(value["time"])
        if timestamp not in idlists:
            idlists[timestamp] = []
        idlists[timestamp] += [value["id"]]

    datasets = {}
    for timestamp, idlist in idlists.items():

        if len(idlist) == 0:
            print("No data associated with timestamp '" + timestamp + "'.")
            continue

        dataset = {"title": timestamp, "data_type": "MS"}

        metadatas = dict([(i, cinfd.get_metadata(i)) for i in idlist])

        unixtimes = [metadatas[i]["unixtime"] for i in idlist]
        if len(set(unixtimes)) > 1:
            msg = "unix times don't match for timestamp '" + timestamp + "'!"
            raise ValueError(msg)

        dataset["tstamp"] = unixtimes[0]
        dataset["timestamp"] = metadatas[idlist[0]]["time"].strftime("%H:%M:%S")

        labels = [metadatas[i]["mass_label"] for i in idlist]
        if "Mass Scan" in labels:
            dataset["scan_type"] = "mass"
        else:
            dataset["scan_type"] = "time"

        dataset["data_cols"] = set()
        dataset["timecols"] = {}
        for i in idlist:  # avoiding id since it's got a builtin meaning
            data = cinfd.get_data(i)
            label = metadatas[i]["mass_label"]
            if len(data.shape) == 1:
                dataset[label] = data
                dataset["data_cols"].add(label)
            elif data.shape[1] == 2:
                x = data[:, 0]
                y = data[:, 1]
                x_label = label + "-x"
                y_label = label + "-y"
                dataset["timecols"][y_label] = x_label  # Fixed 20B26!!!
                dataset[x_label] = x * 1e-3  # cinfdata saves time in ms!!!
                dataset[y_label] = y

                dataset["data_cols"].add(x_label)
                dataset["data_cols"].add(y_label)

        datasets[timestamp] = dataset

    timescans = [
        dataset for dataset in datasets.values() if dataset["scan_type"] == "time"
    ]

    combined = synchronize(timescans, t_zero="first")

    return combined


def get_xy(
    data, xcol=None, ycol=None, label=None,
):
    """
    """
    if xcol is None:
        xcol = label + "-x"
    if ycol is None:
        ycol = label + "-y"
    return data[xcol], data[ycol]


def set_xy(
    data, x, y, xcol=None, ycol=None, label=None,
):
    """
    """
    if xcol is None:
        xcol = label + "-x"
    if ycol is None:
        ycol = label + "-y"
    data[xcol] = x
    data[ycol] = y


def remove_repeats(data, xcol=None, ycol=None, label=None):
    """
    """
    x_0, y_0 = get_xy(data, xcol, ycol, label)
    x, y = only_while_increasing(x_0, y_0)
    set_xy(data, x=x, y=y, xcol=xcol, ycol=ycol, label=label)


def only_while_increasing(x=None, y=None):
    """
    removes the repeats if a dataset goes back and repeats, as happens in
    the Analog In anomoly first observed 18D05.
    Does so in a vectorized way (only loops over "cliff points" where x falls)
    x is monotonically increasing in the returned data
    """

    x_up = np.append(x[1:], x[-1] + 1)  # x shifted up one, so that x_up[i] = x[i+1]
    cliff_points = np.where(x_up < x)[0]  # points right before a drop in x
    mask = np.tile(True, np.size(x))

    for point in cliff_points:
        x_cliff = x[point]
        mask[point:] = np.logical_and(mask[point:], x[point:] > x_cliff)

    return x[mask], y[mask]


def import_set(
    directory,
    MS_file="QMS.txt",
    MS_data=None,
    t_zero="start",
    EC_file=None,
    tag="01",
    cutit=False,
    cut_buffer=60,
    verbose=True,
    override=False,
):
    from .Combining import synchronize, sort_time

    if verbose:
        print("\n\nfunction import_set at your service!\n")

    lslist = os.listdir(directory)

    if MS_data is None:
        if type(MS_file) is str:
            MS_file = [MS_file]
        MS_datas = [
            load_from_file(directory + os.sep + f, data_type="MS", verbose=verbose)
            for f in MS_file
        ]
        MS_data = synchronize(MS_datas, verbose=verbose)
        if len(MS_datas) > 1:
            sort_time(MS_data)

    if EC_file is None:
        EC_file = [f for f in lslist if f[:2] == tag and f[-4:] == ".mpt"]
    elif type(EC_file) is str:
        EC_file = [EC_file]
    EC_datas = [
        load_from_file(directory + os.sep + f, verbose=verbose, data_type="EC")
        for f in EC_file
    ]
    EC_data = synchronize(EC_datas, verbose=verbose)
    if "loop number" in EC_data["data_cols"]:
        sort_time(EC_data, verbose=verbose)  # note, sort_time no longer returns!

    data = synchronize(
        [MS_data, EC_data],
        t_zero=t_zero,
        verbose=verbose,
        override=override,
        cutit=cutit,
        cut_buffer=cut_buffer,
    )
    if verbose:
        print("\nfunction import_set finished!\n\n")
    return data


def save_as_text(
    filename,
    dataset,
    cols="all",
    mols=[],
    tspan="all",
    header=None,
    N_chars=None,
    timecols={},
    **kwargs,
):
    """
    kwargs is fed directly to Molecule.get_flux()
    """
    from .Combining import get_timecol, cut

    lines = []
    if type(header) is list:
        lines += header
    elif type(header) is str:
        lines += [header]

    if cols == "all":
        cols = list(dataset["data_cols"])
    if N_chars is None:
        N_chars = max([len(col) for col in cols])

    col_header = ""
    i_col = 0
    columns = []
    datas = {}
    for col in cols:
        if col in timecols:
            tcol = timecols[col]
        else:
            tcol = get_timecol(col)
        if tcol in dataset and tcol not in columns:  # don't want same tcol twice
            col_header += ("{0:>" + str(N_chars) + "s},\t").format(tcol)
            columns += [tcol]
            i_col += 1
        if col in dataset and col not in columns:  # don't want same tcol twice
            col_header += ("{0:>" + str(N_chars) + "s},\t").format(col)
            columns += [col]
            i_col += 1
        else:
            print(col + " not in dataset. ignoring it.")
            continue
        if tcol in columns:
            x, y = dataset[tcol].copy(), dataset[col].copy()
            if tspan is not False and not tspan == "all":
                x, y = cut(x, y, tspan=tspan)
            datas[tcol], datas[col] = x, y
        else:
            print(
                "timecol '"
                + tcol
                + "' for col '"
                + col
                + "' is not in dataset, so can't cut it."
            )
            datas[col] = dataset[col].copy()

    for mol in mols:
        tcol = mol.name + "_" + mol.primary + "-x"
        col = mol.name + "_" + mol.primary + "-y"
        x, y = mol.get_flux(dataset, tspan=tspan, **kwargs)
        datas[tcol] = x
        datas[col] = y
        col_header += ("{0:>" + str(N_chars) + "s},\t").format(tcol)
        columns += [tcol]
        col_header += ("{0:>" + str(N_chars) + "s},\t").format(col)
        columns += [col]

    lines += [col_header + "\n"]

    i_data = 0
    finished = False
    while not finished:
        N_unfinished = 0
        line = ""
        for col in columns:
            try:
                d = datas[col][i_data]
                line += ("{0:>" + str(N_chars) + ".6g},\t").format(d)
                N_unfinished += 1
            except IndexError:
                line += " " * N_chars + ",\t"
        if N_unfinished == 0:
            finished = True
        else:
            lines += [line + "\n"]
            i_data += 1

    with open(filename, "w") as f:
        f.writelines(lines)


def save_results_as_text(name, cols="all", **kwargs):
    if cols == "all":
        cols = list(kwargs.keys())
    header_line = "".join([(col + ", \t") for col in cols])
    header_line = header_line[:-3] + "\n"
    lines = [header_line]

    N = len(kwargs[cols[0]])

    for i in range(N):
        l = ""
        for col in cols:
            try:
                v = kwargs[col][i]
            except IndexError:
                s = ", \t"
            else:
                s = "{:6.4g}".format(v) + ", \t"
            l += s
        l = l[:-3] + "\n"
        lines += [l]

    with open(name, "w") as f:
        f.writelines(lines)


def import_folder(directory, tags=None, MS_file=None, verbose=True):
    """
    Copined 19G24 from commit  Submitting Trimarco2018 e12b8e9 on Dec 19, 2017

    import everything you need from directory at once.
    tags = None imports as one dataset
    tags = 'all' separates by EC file tag, imports evertything
    tags = ['01','02'] imports two datasets, one for each tag
    """
    from .Combining import synchronize

    if verbose:
        print("\n\nfunction 'imoprt_folder' at your service!\n")
        print("Importing from '" + directory + "'")

    if directory[-1] == os.sep:
        directory = directory[:-1]  # import_set adds the os.sep

    lslist = os.listdir(directory)

    if MS_file is None:
        MS_file = [f for f in lslist if "QMS" in f]

    # importing MS data here rather than in import_set to avoid redundant importing
    MS_datas = [
        import_data(directory + os.sep + f, data_type="MS", verbose=verbose)
        for f in MS_file
    ]
    MS_data = synchronize(MS_datas, verbose=verbose)
    # if len(MS_datas) > 1:          #not necessary, as synchronize sorts by recstart
    # sort_time(MS_data)

    if tags is None:
        EC_file = [f for f in lslist if ".mpt" in f]
        Datasets = import_set(
            directory, MS_data=MS_data, EC_file=EC_file, verbose=verbose
        )
    # sort_time(Datasets)  #probably not necessary, as synchronize sorts by recstart

    else:
        if tags == "all":
            taglist = {f[0:2] for f in lslist if f[-4:] == ".mpt"}
        else:
            taglist = tags  # Renamed so I keep info on what tags originally was
        Datasets = dict(
            [
                (t, import_set(directory, MS_data=MS_data, tag=t, verbose=verbose))
                for t in taglist
            ]
        )

    if verbose:
        print("\n\nfunction 'imoprt_folder' finished!\n")
    return Datasets


def download_data(
    IDs="today",
    timestamps=None,
    data_type="fullscan",
    timestamp_interval=None,
    comment=None,
    connect={},
    verbose=True,
):
    """
    Copied 19G25 from commit   17G28 better combining and plotting b70b43b on Jul 28, 2017
    ... but it seems to be broken :( Not wasting time on this now.


    Returns data columns matching a certain set of ID's.

    """
    import sys

    try:
        import MySQLdb  # known to pip as mysqlclient

        # CONNECT_EXCEPTION = MySQLdb.OperationalError
        # LOG.info('Using MySQLdb as the database module')
        print("imported MySQLdb no problem!")
    except ImportError:
        try:
            # .. if that fails, try with pymysql
            import pymysql as MySQLdb

            MySQLdb.install_as_MySQLdb()
            # CONNECT_EXCEPTION = MySQLdb.err.OperationalError
            # LOG.info('Using pymysql as the database module')
            if sys.version_info.major > 2:
                # LOG.info('pymysql is known to be broken with Python 3. Consider '
                #        'installing mysqlclient!')
                pass
        except:
            print("Error, can't connect to database!")

    connect_0 = dict(
        host="servcinf-sql",  # your host, usually localhost, servcinf would also work, but is slower (IPv6)
        #    port=9995,  # your forwording port
        user="cinf_reader",  # your username
        passwd="cinf_reader",  # your password
        db="cinfdata",
    )  # name of the data base

    for key, val in connect_0.items():
        if key not in connect:
            connect[key] = val

    if data_type == "fullscan":
        data_string_template = (
            "SELECT x,y FROM xy_values_sniffer where measurement = {0} order by id"
        )

    #       try:
    print("Connecting to CINF database...")
    cnxn = MySQLdb.connect(**connect)
    cursor = cnxn.cursor()
    print("Connection successful!")
    #      except:
    #         print('Connection failed!')

    if type(IDs) is int:
        IDs = [IDs]

    datasets = {}
    for ID in IDs:
        data_string = data_string_template.format(str(ID))
        cursor.execute(data_string)
        raw_data = cursor.fetchall()
        list_data = np.array(raw_data)
        xy_data = np.swapaxes(list_data, 0, 1)
        datasets[ID] = xy_data

    return datasets